An intelligent power distribution monitoring system and method

Abstract

The application relates to the technical field of data monitoring, and discloses an intelligent power distribution monitoring system and method, which comprises the following steps: collecting current and voltage signals of preset points to establish a signal data chain; pre-processing the signal data chain based on a window function to obtain a windowed data chain, determining an overlap ratio for window division according to the length of the windowed data chain, and obtaining a plurality of spectrums; comparing the spectrums two by two, and judging whether to perform abnormal early warning according to the comparison result; when early warning is determined, processing each spectrum based on Hilbert transformation to obtain an instantaneous envelope and an instantaneous phase, and obtaining the phase difference value of two same-frequency harmonics in all spectrums; and taking the phase difference value and the spectral similarity as characteristic data, comparing the characteristic data with historical early warning data sets, and performing early warning. The application overcomes the problems of traditional methods, such as dependence on fixed threshold values, slow response and high false alarm rate, and improves the identification accuracy and response timeliness of dynamic harmonic abnormalities.

Description

Technical Field

[0001] This invention relates to the field of data monitoring technology, and more specifically, to an intelligent power distribution monitoring system and method. Background Technology

[0002] With the integration of new energy sources, the widespread application of power electronic equipment, and the diversification of user loads, harmonic pollution in power distribution systems is becoming increasingly prominent, seriously affecting power quality, safe operation, and equipment lifespan. To improve the intelligent operation and maintenance capabilities of power distribution systems, the power system urgently needs to construct intelligent power distribution monitoring methods with real-time monitoring, anomaly identification, and fault early warning capabilities.

[0003] However, traditional FFT methods are weak in processing non-stationary or perturbed signals, making it difficult to accurately identify dynamic harmonic anomalies. Direct FFT processing of unwindowed signals easily leads to spectral leakage, affecting the accuracy of amplitude and phase measurements. Furthermore, current methods mostly focus on harmonic amplitude characteristics, failing to effectively explore the evolution of phase differences between voltage and current signals, resulting in inaccurate determination of harmonic source direction and propagation path. Moreover, relying on fixed threshold judgment mechanisms lacks the ability to learn from and adapt to historical data, easily leading to false alarms and missed alarms.

[0004] Therefore, it is necessary to design an intelligent power distribution monitoring system and method to solve the problems existing in the current technology. Summary of the Invention

[0005] In view of this, the present invention proposes an intelligent power distribution monitoring system and method, which aims to solve the problems of difficulty in capturing time-varying characteristics, spectrum leakage and aliasing, low utilization of phase information, and unintelligent early warning mechanism.

[0006] In one aspect, the present invention proposes an intelligent power distribution monitoring method, comprising:

[0007] A signal data chain is established by collecting current and voltage signals from preset points, including distribution transformers and user incoming lines. The signal data chain includes several acquisition time points and several current and voltage signals, and each acquisition time point corresponds to a current and voltage signal.

[0008] The signal data chain is preprocessed based on the windowing function to obtain the windowed data chain. The overlap ratio is determined according to the length of the windowed data chain to divide the window and obtain several spectra.

[0009] The spectrum is compared pairwise, and an abnormality warning is determined based on the comparison results. When an abnormality warning is determined, each spectrum is processed based on Hilbert transform to obtain the instantaneous envelope and instantaneous phase, and the phase difference between two harmonics of the same frequency in all the spectrum is obtained.

[0010] The phase difference and spectral similarity are used as feature data. The feature data are compared with the historical early warning dataset, and an early warning is issued based on the comparison results.

[0011] Furthermore, when establishing a signal data link by acquiring current and voltage signals from preset points, the process includes:

[0012] Perform wavelet transform on each of the current and voltage signals to extract high-frequency signals;

[0013] The current and voltage signals were decomposed into 6 levels of wavelet using Daubechies 6 wavelet discrete wavelet transform.

[0014] The soft thresholding method is used to extract detail signals, with the threshold set to 1.5 times the standard deviation of the signal.

[0015] The extracted signal is then subjected to inverse wavelet transform to establish the signal data link.

[0016] Furthermore, when the signal data chain is preprocessed based on a windowing function to obtain the windowed data chain, the windowing function includes a Hanning window or a Hamming window.

[0017] Furthermore, when dividing the window to obtain several spectra by determining the overlap ratio based on the length of the data link after windowing, the following are included:

[0018] The length of the windowed data chain is compared with a preset length threshold, and the overlap ratio is determined based on the comparison result. The overlap ratio is directly proportional to the length of the windowed data chain.

[0019] Furthermore, when comparing the spectrum pairwise and determining whether to issue an anomaly warning based on the comparison results, the process includes:

[0020] The spectral similarity between any two spectra is obtained based on Euclidean distance. The spectral similarity is then compared with a similarity threshold, and an abnormality warning is issued based on the comparison result.

[0021] When the spectral similarity is less than the similarity threshold, an anomaly warning is issued.

[0022] When the spectral similarity is greater than or equal to the similarity threshold, it is determined that no abnormality warning will be issued.

[0023] Furthermore, when processing each spectrum based on the Hilbert transform to obtain the instantaneous envelope and instantaneous phase, the process includes:

[0024]

[0025] Where z(t) represents the analytic signal and x(t) represents the original real signal. Let A(t) represent the signal after the Hilbert transform, which is the orthogonal component of x(t), where A(t) represents the instantaneous envelope and φ(t) represents the instantaneous phase.

[0026] Furthermore, when comparing the feature data with historical early warning datasets and issuing early warnings based on the comparison results, the process includes:

[0027] The historical early warning dataset includes several historical feature data and several historical early warning levels, and each of the historical feature data corresponds to a historical early warning level.

[0028] Initialize K centroids from all the historical feature data, and assign the feature data to the nearest centroid to form K clusters; repeat the process of assigning centroids and calculating the centroids of each cluster after assignment until the centroids no longer change;

[0029] When historical feature data exists in a cluster containing the feature data, the historical features in that cluster are taken as a similarity set, and the historical warning level corresponding to the maximum feature similarity in the similarity set is determined for warning purposes.

[0030] When historical feature data is absent in a cluster containing the aforementioned feature data, an early warning is issued based on the prediction results of the long-term and short-term prediction models.

[0031] Furthermore, when issuing early warnings based on the prediction results of long-term and short-term forecasting models, this includes:

[0032] The historical early warning dataset is sampled according to a preset ratio to obtain a training subset and a test subset;

[0033] The LSTM long short-term prediction model is iteratively trained using the training subset, and the LSTM long short-term prediction model after iterative training is evaluated based on the test subset to obtain the final LSTM long short-term prediction model.

[0034] The prediction results are obtained by inputting the feature data into the trained prediction model.

[0035] Compared with existing technologies, the advantages of this invention are as follows: By synchronously collecting current and voltage signals at preset key points such as distribution transformers and user incoming lines, a time-continuous signal data chain is established, enabling a comprehensive understanding of the propagation characteristics of harmonics in the distribution network; by introducing windowing functions and overlapping averaging preprocessing methods, spectral leakage and boundary effects are effectively suppressed, improving the accuracy and stability of spectral analysis; by using Hilbert transform to extract the instantaneous envelope and phase information of the signal, the phase difference between harmonic signals of the same frequency can be accurately calculated, thereby identifying the direction of the harmonic source and the propagation path; by using spectral similarity and phase difference as core features and comparing them with historical early warning datasets, an adaptive anomaly identification and early warning mechanism is realized, overcoming the problems of traditional methods relying on fixed thresholds, slow response, and high false alarm rates, and improving the accuracy and timeliness of dynamic harmonic anomaly identification.

[0036] On the other hand, this application also provides an intelligent power distribution monitoring system for applying the above-mentioned intelligent power distribution monitoring method, including:

[0037] The acquisition unit is configured to acquire current and voltage signals at preset points to establish a signal data link. The preset points include distribution transformers and user incoming lines. The signal data link includes several acquisition time points and several current and voltage signals, and each acquisition time point corresponds to a current and voltage signal.

[0038] The processing unit is configured to preprocess the signal data chain based on a windowing function to obtain a windowed data chain, and to divide the window according to the overlap ratio determined by the length of the windowed data chain to obtain several spectra.

[0039] The judgment unit is configured to compare the spectrum pairwise and determine whether to issue an abnormal warning based on the comparison result; when it is determined to issue a warning, it processes each spectrum based on Hilbert transform to obtain the instantaneous envelope and instantaneous phase, and obtains the phase difference value of two harmonics of the same frequency in all the spectrum.

[0040] The early warning unit is configured to use the phase difference value and spectral similarity as feature data, compare the feature data with the historical early warning dataset, and issue an early warning based on the comparison result.

[0041] Furthermore, it also includes a storage unit configured to store the feature data and the prediction results.

[0042] It is understandable that the aforementioned intelligent power distribution monitoring systems and methods have the same beneficial effects, and will not be elaborated upon here. Attached Figure Description

[0043] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0044] Figure 1 A flowchart of the intelligent power distribution monitoring method provided in the embodiments of the present invention;

[0045] Figure 2 This is a functional block diagram of an intelligent power distribution monitoring system provided in an embodiment of the present invention. Detailed Implementation

[0046] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0047] In some embodiments of this application, see Figure 1 As shown, an intelligent power distribution monitoring method includes:

[0048] S100: Collect current and voltage signals from preset points to establish a signal data link. The preset points include distribution transformers and user incoming lines. The signal data link includes several acquisition time points and several current and voltage signals, and each acquisition time point corresponds to a current and voltage signal.

[0049] S200: The signal data link is preprocessed based on the windowing function to obtain the windowed data link. The overlap ratio is determined according to the length of the windowed data link to divide the window and obtain several spectra.

[0050] S300: Compares the spectrum pairwise and determines whether to issue an anomaly warning based on the comparison results. When an anomaly warning is determined, each spectrum is processed based on the Hilbert transform to obtain the instantaneous envelope and instantaneous phase, and the phase difference between two harmonics of the same frequency in all spectra is obtained.

[0051] S400: The phase difference and spectral similarity are used as feature data. The feature data are compared with the historical warning dataset, and warnings are issued based on the comparison results.

[0052] Specifically, sampling points are first set up at key nodes such as distribution transformers and user incoming lines to collect current and voltage signals at various times, constructing a signal data chain containing time series data to obtain the spatial and temporal propagation characteristics of middle harmonics. Windowing functions (such as Hamming and Hanning windows) are used to preprocess the signal data chain, reducing spectral leakage while maintaining frequency resolution. The windowed data chain is further divided into windows by setting an overlap ratio to improve spectral continuity and accuracy. After spectral analysis of the signals within each window, pairwise spectral comparisons are performed. By comparing the spectral structure and energy distribution, potential anomalies are initially identified. When an abnormal trend is determined, a Hilbert transform is applied to the spectral signal to extract the instantaneous envelope and instantaneous phase of each frequency component. The phase difference between key frequency harmonics is then calculated to infer the source of the harmonics and the path of the anomaly. The obtained phase difference and spectral similarity are used as a combined feature vector and compared with historical early warning datasets to achieve an adaptive early warning judgment mechanism that integrates historical experience, thereby improving the accuracy and responsiveness of early warnings.

[0053] Understandably, a robust intelligent power distribution monitoring mechanism is constructed through multi-dimensional fusion of time and frequency domains, amplitude and phase, and current and historical data. Compared to the traditional passive identification method that relies solely on amplitude thresholds, the introduction of windowed spectrum processing and phase difference discrimination mechanisms enhances the ability to perceive the dynamic evolution characteristics of harmonics. Furthermore, through a data-driven adaptive comparison algorithm, accurate identification and timely early warning of abnormal events are achieved.

[0054] In some embodiments of this application, the process of acquiring current and voltage signals from preset points to establish a signal data link includes: performing wavelet transform on each current and voltage signal to extract high-frequency signals; using the Daubechies6 wavelet for discrete wavelet transform to decompose the current and voltage signals into 6 levels of wavelet decomposition; using a soft thresholding method to extract detail signals, with the threshold set to 1.5 times the signal standard deviation; and establishing the signal data link after performing inverse wavelet transform on the extracted signals.

[0055] Specifically, the Daubechies 6 (db6) wavelet function, known for its good local characteristics and stationarity, is selected as the mother wavelet. A six-level wavelet decomposition is performed on each sampled signal, breaking down the original signal into a low-frequency approximate component (A6) and multiple high-frequency detail components (D1–D6). The high-frequency components are more sensitive to anomalies such as harmonics and impulse interference. Based on this, a soft-threshold denoising method is used to process the detail signals in order to suppress noise and preserve high-frequency anomaly characteristics. A threshold is set, equal to 1.5 times the standard deviation of the detail signal (i.e., λ = 1.5σ). Detail components with amplitudes below the threshold are directly attenuated to zero, while those above the threshold are smoothly reduced, preserving their main energy and reducing sharp transitions. The soft-threshold denoising strategy balances noise reduction and signal integrity, outperforming hard-threshold denoising in terms of continuity and smoothness. After processing, an inverse wavelet transform is performed on the denoised signal to reconstruct the denoised voltage and current signals. The resulting signal data chain has a higher signal-to-noise ratio and dynamic anomaly response capability, providing a cleaner and more sensitive signal input for subsequent processing steps such as windowing, spectrum analysis, and Hilbert transform.

[0056] Understandably, leveraging the multi-scale characteristics of wavelets enables precise extraction and noise suppression of high-frequency anomalies in power signals, effectively enhancing the ability to express information related to harmonic anomalies in the original signal. Through the advantages of the db6 wavelet in edge detection and harmonic capture, and the smoothness ensured by soft thresholding during signal reconstruction, the established signal data chain not only retains key feature information but also significantly reduces the interference of high-frequency noise on subsequent analysis, thereby improving the stability and accuracy of the entire intelligent power distribution monitoring system.

[0057] In some embodiments of this application, when the signal data link is preprocessed based on a windowing function to obtain the windowed data link, the windowing function includes a Hanning window or a Hamming window.

[0058] In some embodiments of this application, when dividing the window to obtain several spectra by determining the overlap ratio based on the windowed data link length, the method includes: comparing the windowed data link length with a preset length threshold, determining the overlap ratio based on the comparison result, wherein the overlap ratio is proportional to the windowed data link length.

[0059] Specifically, the Hanning window: Hamming Window: 'n' represents the sampling point index within the window, and 'N' represents the window length. The window function effectively eliminates abrupt changes caused by truncation by weighted attenuation at both ends of the signal, thereby improving the accuracy of spectral energy distribution. The windowed data chain is further divided into multiple analysis windows for spectral transformation. During window division, a strategy of adaptively adjusting the overlap ratio based on the data chain length is introduced. This involves comparing the current windowed data chain length with a preset length threshold. If the data chain is long, the overlap ratio is increased to enhance the temporal continuity and spectral smoothness of the analysis; if the data chain is short, the overlap ratio is appropriately reduced to ensure frequency resolution and computational efficiency. This achieves a dynamic adjustment mechanism where the overlap ratio is proportional to the data chain length, maintaining the accuracy of spectral estimation while avoiding redundant calculations caused by excessive overlap.

[0060] Understandably, selecting a window function mitigates the impact of spectral leakage, while dynamically setting the overlap ratio based on signal length enhances the flexibility and adaptability of spectral analysis. Compared to traditional methods with fixed window lengths and overlap ratios, this approach intelligently adjusts processing parameters based on the amount of data acquired in different operating scenarios, making the spectrum extraction process more robust and efficient.

[0061] In some embodiments of this application, when comparing spectrum pairs and determining whether to issue an anomaly warning based on the comparison results, the process includes: obtaining the spectral similarity between each pair of spectra based on Euclidean distance, comparing the spectral similarity with a similarity threshold, and determining whether to issue an anomaly warning based on the comparison results. When the spectral similarity is less than the similarity threshold, an anomaly warning is issued. When the spectral similarity is greater than or equal to the similarity threshold, no anomaly warning is issued.

[0062] Understandably, the similarity of the spectra is then deduced from the distance. The smaller the distance, the closer the two spectra are, and the higher the similarity. Conversely, a larger distance indicates a change in the spectra structure, potentially indicating power quality anomalies or harmonic disturbances. During the comparison, the obtained spectra similarity is compared with a preset similarity threshold: if the similarity is below the threshold (i.e., the spectra differ significantly), it is determined that the current spectrum shows a significant deviation from historical or normal conditions, triggering an alert; if the similarity is greater than or equal to the threshold (the spectra structure has not changed significantly), it is considered to be operating normally, and no alert is issued. By judging the relative change in the global characteristics of the spectra shape, system stability is determined, avoiding the risk of misjudgment based solely on single-frequency amplitude or instantaneous jumps.

[0063] In some embodiments of this application, when processing each spectrum based on the Hilbert transform to obtain the instantaneous envelope and instantaneous phase, the following steps are included:

[0064]

[0065] Where z(t) represents the analytic signal and x(t) represents the original real signal. Let A(t) represent the signal after the Hilbert transform, which is the orthogonal component of x(t), where A(t) represents the instantaneous envelope and φ(t) represents the instantaneous phase.

[0066] It is understandable that j represents the imaginary unit. By using the Hilbert transform to extract the instantaneous envelope and instantaneous phase of corresponding harmonics from various frequencies in the spectrum, the phase difference of the same harmonic at different monitoring points can be accurately calculated. This is crucial for analyzing the direction of harmonic sources, identifying system disturbances, and determining abnormal signal trends. By constructing an analytical signal using the Hilbert transform and extracting the signal's envelope and phase as instantaneous features, compared to static frequency domain features (such as amplitude and energy), it can reflect the dynamic evolution of the signal over time. It exhibits higher sensitivity and time-based positioning accuracy, especially for transient events and non-stationary disturbances, further enhancing the intelligent power distribution monitoring system's ability to analyze harmonic dynamic behavior and its early warning accuracy.

[0067] In some embodiments of this application, when comparing feature data with a historical early warning dataset and issuing an early warning based on the comparison result, the following steps are taken: The historical early warning dataset includes several historical feature data and several historical early warning levels, and each historical feature data corresponds to a historical early warning level. K centroids are initialized from all historical feature data, and feature data are assigned to the nearest centroid to form K clusters. The centroid of each cluster is recalculated. The process of assigning centroids and recalculating the centroids of each assigned cluster is repeated until the centroids no longer change.

[0068] Specifically, when historical feature data exists within a cluster containing feature data, the historical features in that cluster are treated as a similarity set, and the historical warning level corresponding to the maximum feature similarity in the similarity set is determined for issuing a warning. When historical feature data does not exist within a cluster containing feature data, a warning is issued based on the prediction results of the long-term and short-term prediction models.

[0069] In some embodiments of this application, when issuing early warnings based on the prediction results of a long-term and short-term prediction model (LSTM), the process includes: sampling historical early warning datasets according to a preset ratio to obtain a training subset and a test subset; iteratively training the LSTM using the training subset; evaluating the iteratively trained LSTM using the test subset to obtain the final LSTM; and inputting feature data into the trained prediction model to obtain prediction results.

[0070] Specifically, several sets of historical feature data are used, each composed of multi-dimensional values ​​such as "phase difference + spectral similarity"; the corresponding historical warning levels (e.g., normal, slight anomaly, severe anomaly) serve as label values. K-means uses unsupervised learning to mine the inherent distribution structure in historical data, achieving automatic clustering and similarity matching without thresholds; LSTM, based on time series learning capabilities, can capture the evolution trend of feature values ​​over time and perform dynamic prediction and completion. This ensures both the timeliness of warnings and improves the ability to infer unseen scenarios.

[0071] Understandably, by combining K-means and LSTM machine learning techniques, high-dimensional intelligent discrimination of spectral and phase features in voltage and current signals is achieved. This approach is independent of fixed thresholds and can adapt to various operating conditions and equipment states. Furthermore, the introduction of historical experience data for analogical enhancement effectively improves the accuracy and stability of early warning judgments, thereby enhancing the power distribution system's ability to monitor and warn of complex nonlinear harmonic disturbances.

[0072] In the above embodiments, by synchronously acquiring current and voltage signals at preset key points such as distribution transformers and user incoming lines, a time-continuous signal data chain is established, enabling a comprehensive understanding of the propagation characteristics of harmonics in the distribution network. By introducing windowing functions and overlapping averaging preprocessing methods, spectral leakage and boundary effects are effectively suppressed, improving the accuracy and stability of spectral analysis. Hilbert transform is used to extract the instantaneous envelope and phase information of the signal, further enabling accurate calculation of the phase difference between harmonic signals of the same frequency, thus allowing identification of the harmonic source direction and propagation path. Using spectral similarity and phase difference as core features, combined with comparison using historical early warning datasets, an adaptive anomaly identification and early warning mechanism is realized, overcoming the problems of traditional methods relying on fixed thresholds, slow response, and high false alarm rates, thus improving the accuracy and timeliness of dynamic harmonic anomaly identification.

[0073] In another preferred embodiment based on the above embodiments, see [reference] Figure 2 As shown, this embodiment provides an intelligent power distribution monitoring system for applying the above-described intelligent power distribution monitoring method, including:

[0074] The acquisition unit is configured to acquire current and voltage signals at preset points to establish a signal data link. The preset points include distribution transformers and user incoming lines. The signal data link includes several acquisition time points and several current and voltage signals, with each acquisition time point corresponding to a current and voltage signal.

[0075] The processing unit is configured to preprocess the signal data link based on the windowing function to obtain the windowed data link, and divide the window according to the overlap ratio determined by the length of the windowed data link to obtain several spectra.

[0076] The judgment unit is configured to compare the spectrum pairwise and determine whether to issue an anomaly warning based on the comparison results. When an anomaly warning is determined, each spectrum is processed based on the Hilbert transform to obtain the instantaneous envelope and instantaneous phase, and the phase difference between two harmonics of the same frequency in all spectra is obtained.

[0077] The early warning unit is configured to use phase difference and spectral similarity as feature data, compare the feature data with historical early warning datasets, and issue early warnings based on the comparison results.

[0078] Furthermore, it also includes storage units configured to store feature data and prediction results.

[0079] Understandably, by synchronously collecting current and voltage signals at preset key points such as distribution transformers and user incoming lines, and establishing a time-continuous signal data chain, it is possible to comprehensively understand the propagation characteristics of harmonics in the distribution network. By introducing windowing functions and overlapping averaging preprocessing methods, spectral leakage and boundary effects are effectively suppressed, improving the accuracy and stability of spectral analysis. Hilbert transform is used to extract the instantaneous envelope and phase information of the signal, further enabling accurate calculation of the phase difference between harmonic signals of the same frequency, thus allowing identification of the harmonic source direction and propagation path. Using spectral similarity and phase difference as core features, combined with comparison with historical early warning datasets, an adaptive anomaly identification and early warning mechanism is realized, overcoming the problems of traditional methods relying on fixed thresholds, slow response, and high false alarm rates, thereby improving the accuracy and timeliness of dynamic harmonic anomaly identification.

[0080] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program goods. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program goods embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0081] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program goods according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1A device that provides the functions specified in one or more boxes.

[0082] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0083] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0084] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method of intelligent power distribution monitoring, characterized by, include: A signal data chain is established by collecting current and voltage signals from preset points, including distribution transformers and user incoming lines. The signal data chain includes several acquisition time points and several current and voltage signals, and each acquisition time point corresponds to a current and voltage signal. The signal data chain is preprocessed based on the windowing function to obtain the windowed data chain. The overlap ratio is determined according to the length of the windowed data chain to divide the window and obtain several spectra. The spectrum is compared pairwise, and an abnormality warning is issued based on the comparison results. When a warning is issued, the instantaneous envelope and instantaneous phase are obtained by processing each spectrum based on the Hilbert transform, and the phase difference between two harmonics of the same frequency in all the spectrums is obtained. The phase difference and spectral similarity are used as feature data. The feature data are compared with the historical early warning dataset, and an early warning is issued based on the comparison results. When comparing the spectrum pairwise and determining whether to issue an anomaly warning based on the comparison results, the process includes: The spectral similarity between any two spectra is obtained based on Euclidean distance. The spectral similarity is then compared with a similarity threshold, and an abnormality warning is issued based on the comparison result. When the spectral similarity is less than the similarity threshold, an anomaly warning is issued. When the spectral similarity is greater than or equal to the similarity threshold, it is determined that no abnormality warning will be issued.

2. The intelligent power distribution monitoring method of claim 1, wherein, When establishing a signal data link by acquiring current and voltage signals at preset points, the following steps are included: Perform wavelet transform on each of the current and voltage signals to extract high-frequency signals; The current and voltage signals were decomposed into 6 levels of wavelet using Daubechies 6 wavelet discrete wavelet transform. The soft thresholding method is used to extract detail signals, with the threshold set to 1.5 times the standard deviation of the signal. The extracted signal is then subjected to inverse wavelet transform to establish the signal data link.

3. The intelligent power distribution monitoring method of claim 1, wherein, When the signal data link is preprocessed based on a windowing function to obtain the windowed data link, the windowing function includes a Hanning window or a Hamming window.

4. The intelligent power distribution monitoring method of claim 3, wherein, When dividing the window based on the overlap ratio determined by the length of the data link after windowing, and obtaining several spectra, including: The length of the windowed data chain is compared with a preset length threshold, and the overlap ratio is determined based on the comparison result. The overlap ratio is directly proportional to the length of the windowed data chain.

5. The intelligent power distribution monitoring method of claim 1, wherein, When processing each spectrum using the Hilbert transform to obtain the instantaneous envelope and instantaneous phase, the following steps are included: ； wherein represents the analytic signal, represents the original real signal, represents the signal after Hilbert transform, is the quadrature component of represents the instantaneous envelope, represents the instantaneous phase.

6. The intelligent power distribution monitoring method of claim 5, wherein, When comparing the feature data with historical early warning datasets and issuing early warnings based on the comparison results, the process includes: The historical early warning dataset includes several historical feature data and several historical early warning levels, and each of the historical feature data corresponds to a historical early warning level. Initialize K centroids from all the historical feature data, assign the feature data to the nearest centroid to form K clusters; recalculate the centroid of each cluster; repeat the process of assigning centroids and calculating the centroids of each cluster after assignment until the centroids no longer change; When historical feature data exists in a cluster containing the feature data, the historical features in that cluster are taken as a similarity set, and the historical warning level corresponding to the maximum feature similarity in the similarity set is determined for warning purposes. When historical feature data is absent in a cluster containing the aforementioned feature data, an early warning is issued based on the prediction results of the long-term and short-term prediction models.

7. The intelligent power distribution monitoring method of claim 6, wherein, When issuing early warnings based on prediction results from long-term and short-term forecasting models, the following applies: The historical early warning dataset is sampled according to a preset ratio to obtain a training subset and a test subset; The LSTM long short-term prediction model is iteratively trained using the training subset, and the LSTM long short-term prediction model after iterative training is evaluated based on the test subset to obtain the final LSTM long short-term prediction model. The prediction results are obtained by inputting the feature data into the trained prediction model.

8. An intelligent power distribution monitoring system for applying the intelligent power distribution monitoring method according to any one of claims 1 to 7, characterized by, include: The acquisition unit is configured to acquire current and voltage signals at preset points to establish a signal data link. The preset points include distribution transformers and user incoming lines. The signal data link includes several acquisition time points and several current and voltage signals, and each acquisition time point corresponds to a current and voltage signal. The processing unit is configured to preprocess the signal data chain based on a windowing function to obtain a windowed data chain, and to divide the window according to the overlap ratio determined by the length of the windowed data chain to obtain several spectra. The judgment unit is configured to compare the spectrum pairwise and determine whether to issue an anomaly warning based on the comparison results. When a warning is issued, the instantaneous envelope and instantaneous phase are obtained by processing each spectrum based on the Hilbert transform, and the phase difference between two harmonics of the same frequency in all the spectrums is obtained. The early warning unit is configured to use the phase difference value and spectral similarity as feature data, compare the feature data with the historical early warning dataset, and issue an early warning based on the comparison result.

9. The intelligent power distribution monitoring system of claim 8, wherein, Also includes: The storage unit is configured to store the feature data and the prediction results.

Images