A power quality waveform pattern recognition method and system for a distributed power generation system

Abstract

This application proposes a method and system for power quality waveform pattern recognition in distributed generation systems, relating to the field of power quality monitoring and analysis technology for distributed generation systems. The method includes acquiring current waveform data of voltage and current in the power grid, as well as the grid's operating status; establishing and updating a background waveform feature benchmark based on the grid's operating status and the current waveform data; detecting deviations in the current waveform data based on the background waveform feature benchmark to identify one or more potential events of concern; responding to a potential event of concern by backtracking high-resolution waveform data from the time period in which the event occurred; extracting micro-features from the high-resolution waveform data; and attributing the micro-features to the operating status to obtain a judgment result, which includes distinguishing between benign transients, changes caused by component aging, or early signs of faults. This application can improve the accuracy and reliability of identification.

Description

Technical Field

[0001] This application relates to the field of power quality monitoring and analysis technology for distributed generation systems, and more specifically, to a method and system for power quality waveform pattern recognition in distributed generation systems. Background Technology

[0002] With the increasing prevalence of distributed generation systems, the stability and reliability of their power quality are crucial for industrial production and residential life. However, as new energy access methods become more complex and grid operation modes evolve, power quality waveforms often exhibit nonlinear characteristics such as distortion and resonance. Traditional monitoring and identification methods rely on pre-set static pattern libraries and human experience, making it difficult to quickly and accurately identify complex and changing waveform patterns, let alone precisely locate the source of anomalies. This is especially true for early fault precursors that are short-lasting, low-amplitude, and highly intermittent, leading to false alarms and missed alarms, affecting system stability and predictive maintenance capabilities. In distributed generation systems, the introduction of new loads, the implementation of dynamic operation modes, and component aging continuously alter the electrical behavior of the grid. Power electronic equipment, electric vehicle charging piles, and other new nonlinear loads generate waveforms not found in the static pattern library; dynamic strategies such as microgrid islanding switching and demand response produce unique transient and steady-state waveforms that differ significantly from the initial "normal" mode; component aging causes electrical characteristic drift, making the original normal waveform unable to match the current equipment state. These factors lead to continuous changes in the definition of normal and abnormal power quality, i.e., "conceptual drift." Existing technologies rely on statically or periodically updated pattern libraries, causing diagnostic knowledge to quickly become outdated. They cannot distinguish between benign and transient failures, nor can they identify the "new normal" brought about by component aging. Furthermore, they struggle to capture intermittent precursors to early failures, increasing the risk of missed detections and weakening predictive maintenance and proactive fault mitigation capabilities. Manually updating pattern libraries to adapt to all new changes is a complex and unsustainable task. Summary of the Invention

[0003] This application aims to address at least one of the technical problems existing in the prior art. To this end, this application proposes a method and system for identifying power quality waveform patterns in distributed generation systems. The aim is to solve the problems of existing methods for identifying power quality waveform patterns in distributed generation systems, which struggle to quickly and accurately identify complex and constantly changing waveform patterns, accurately locate the source of anomalies, and are prone to false alarms or missed alarms when facing early signs of faults, thus affecting the stable operation and predictive maintenance capabilities of the system.

[0004] In a first aspect, embodiments of this application provide a method for identifying power quality waveform patterns in a distributed generation system, including:

[0005] Acquire the current waveform data of voltage and current in the power grid and the operating status of the power grid, and establish and update the background waveform feature benchmark based on the operating status of the power grid and the current waveform data;

[0006] Based on the current waveform data and the background waveform feature benchmark, deviations in the current waveform data are detected to identify one or more potential events of interest, wherein the potential events of interest characterize events in which deviations occur in the current waveform data;

[0007] In response to the potential event of interest, high-resolution waveform data of the time period in which the potential event of interest occurred is retrieved.

[0008] Micro-features are extracted from the high-resolution waveform data to obtain micro-features, wherein the micro-features are used to characterize the details of the potential event of interest;

[0009] Based on the operating status, the micro-features are attributed and judged to obtain the judgment result, which includes distinguishing between benign transients, changes caused by component aging, or early signs of failure.

[0010] According to some embodiments of this application, establishing and updating the background waveform characteristic reference based on the operating state of the power grid and the current waveform data includes:

[0011] When establishing and updating the background waveform feature benchmark based on the current waveform data of the power grid voltage and current based on the preset monitoring cycle, the current waveform data within each preset monitoring cycle is detected to obtain high-frequency transient energy and waveform distortion measurement.

[0012] When the high-frequency transient energy is detected to exceed a preset threshold, or when the waveform distortion measurement fluctuates within a preset time, the data segment within the preset monitoring period is marked as a potential abnormal data segment.

[0013] When the micro-features of the potential abnormal data segment deviate from the background waveform feature benchmark, and the deviation does not conform to the known benign transient or component aging-induced changes, the potential abnormal data segment is isolated, the isolated potential abnormal data segment is deleted from the update calculation of the background waveform feature benchmark, and the background waveform feature benchmark is established and updated.

[0014] According to some embodiments of this application, establishing and updating the background waveform characteristic reference based on the operating state of the power grid and the current waveform data includes:

[0015] Obtain historical waveform reference values ​​of voltage and current of the power grid during the historical period;

[0016] The preset smoothing factor is adjusted according to the operating status of the power grid to obtain the adjusted preset smoothing factor;

[0017] The background waveform feature benchmark is established and updated by calculating the current waveform data, the historical waveform benchmark value, and the adjusted preset smoothing factor based on the exponentially weighted moving average method.

[0018] According to some embodiments of this application, adjusting the preset smoothing factor according to the operating state of the power grid to obtain the adjusted preset smoothing factor includes:

[0019] When the fluctuation amplitude of the current waveform data is detected to be less than or equal to the first preset threshold, the preset smoothing factor is increased to obtain the adjusted preset smoothing factor.

[0020] When the fluctuation amplitude of the current waveform data is detected to be greater than or equal to the second preset threshold, the preset smoothing factor is reduced to obtain the adjusted preset smoothing factor, wherein the first preset threshold is less than the second preset threshold.

[0021] According to some embodiments of this application, the step of detecting deviations in the current waveform data based on the current waveform data and the background waveform feature benchmark to identify one or more potential events of interest includes:

[0022] Determine the deviation threshold based on the background waveform feature benchmark;

[0023] Establish a statistical control chart for detection deviations;

[0024] The instantaneous value of the current waveform is obtained based on the current waveform data;

[0025] The preset instantaneous value of the waveform is obtained based on the background waveform feature reference;

[0026] The current waveform instantaneous value, the preset waveform instantaneous value, and the deviation threshold are input into the statistical control chart to detect deviations in the current waveform data in order to identify one or more potential events of interest.

[0027] According to some embodiments of this application, the step of retrieving high-resolution waveform data of the time period in which the potential event of concern occurs in response to the potential event of concern includes:

[0028] In response to the potential event of concern, determine the timestamp of the occurrence of the potential event of concern;

[0029] Based on the occurrence timestamp, high-resolution waveform data of the period in which the potential event of concern occurred is traced back.

[0030] According to some embodiments of this application, the micro-feature extraction of the high-resolution waveform data to obtain micro-features, wherein the micro-features are used to characterize the details of the potential event of interest, including:

[0031] Transient wavelet analysis is performed on the high-resolution waveform data to obtain the energy and amplitude in different frequency ranges to obtain micro-features;

[0032] Alternatively, waveform geometric property analysis can be performed on the high-resolution waveform data to obtain the curvature of each sampling point in the high-resolution waveform data in order to obtain micro-features;

[0033] Alternatively, energy spectral density analysis can be performed on the high-resolution waveform data to obtain the energy spectral density of the waveform and thus obtain micro-features.

[0034] According to some embodiments of this application, the step of combining the operating state to perform attribution judgment on the micro-features and obtain the judgment result includes:

[0035] When the operating state is microgrid islanding switching and the micro-feature indicates that the current waveform data has a drop and recovers within a preset time, and the magnitude of the drop is less than the preset magnitude, the judgment result is a benign transient.

[0036] When the operating state is a normal aging trajectory and the micro-feature indicates that the current waveform data has increased, the judgment result is a new normal for component aging.

[0037] When the operating state is an abnormal aging trajectory and not a microgrid islanding switch, and the micro-feature indicates that the deviation of the current waveform data exceeds the preset range, the judgment result is that it is an early sign of an initial fault.

[0038] According to some embodiments of this application, after performing attribution judgment on the micro-features in conjunction with the operating state and obtaining the judgment result, the process includes:

[0039] The judgment result is used to generate an alarm report, wherein the alarm report includes the event time, location, anomaly type and relevant micro-feature values;

[0040] An alarm report, including the event time, location, anomaly type, and relevant micro-feature values, will be sent to the user.

[0041] Secondly, this application also discloses a power quality waveform pattern recognition system for distributed generation systems, comprising:

[0042] The acquisition module is used to acquire the current waveform data of voltage and current in the power grid and the operating status of the power grid, and to establish and update the background waveform feature reference based on the operating status of the power grid and the current waveform data.

[0043] The detection module is used to detect deviations in the current waveform data based on the current waveform data and the background waveform feature benchmark, so as to identify one or more potential events of interest, wherein the potential events of interest characterize events in which deviations occur in the current waveform data;

[0044] The backtracking module is used to respond to the potential event of concern by backtracking the high-resolution waveform data of the time period in which the potential event of concern occurred;

[0045] An extraction module is used to extract micro-features from the high-resolution waveform data to obtain micro-features, wherein the micro-features are used to characterize the details of the potential event of interest;

[0046] The judgment module is used to perform attribution judgment on the micro-features in combination with the operating state and obtain the judgment result. The judgment result includes distinguishing between benign transients, changes caused by component aging, or early signs of failure.

[0047] The technical solution according to the embodiments of this application has at least the following beneficial effects: The power quality waveform pattern recognition method for distributed generation systems disclosed in this application obtains the current waveform data of voltage and current in the power grid and the operating status of the power grid, and establishes and updates the background waveform feature benchmark based on the operating status of the power grid and the current waveform data. This effectively solves the problem that traditional methods rely on static pattern libraries, leading to the rapid obsolescence of diagnostic knowledge, and achieves dynamic adaptation to the continuous evolution of the electrical behavior of the power grid. By detecting deviations between the current waveform data and the background waveform feature benchmark, potential events of concern are identified, enabling timely detection of abnormal situations in the power grid. Furthermore, in response to potential events of concern, high-resolution waveform data is backtracked and micro-feature extraction is performed to obtain detailed information about the event, overcoming the shortcomings of existing technologies in capturing and classifying intermittent, low-amplitude early signs of faults. Ultimately, by combining the operating status to attribute micro-features, the system can distinguish between benign transients, changes caused by component aging, or early signs of failure. This effectively solves the problem that traditional systems cannot distinguish between benign transients and true fault transients under specific operating modes, nor can they accurately identify the "new normal" state of subtle drift caused by component aging. This significantly improves the accuracy and reliability of identification and provides strong support for predictive maintenance and proactive fault mitigation of distributed generation systems.

[0048] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0049] The accompanying drawings are used to provide a further understanding of the technical solutions of this application and constitute a part of the specification. They are used together with the embodiments of this application to explain the technical solutions of this application and do not constitute a limitation on the technical solutions of this application.

[0050] Figure 1 A flowchart illustrating a power quality waveform pattern recognition method for a distributed generation system provided in one embodiment of this application;

[0051] Figure 2 This is a schematic diagram of a power quality waveform pattern recognition system for a distributed generation system provided in one embodiment of this application. Detailed Implementation

[0052] To make the objectives, technical methods, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0053] It should be noted that the meaning of "multiple" (or "more than") in the description of the embodiments of this application refers to two or more, and "greater than," "less than," "exceeding," etc. are understood to exclude the number itself, while "above," "below," "within," etc. are understood to include the number itself. If "first," "second," etc. are used in the description, they are only for the purpose of distinguishing technical features and should not be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated or the order of the technical features indicated.

[0054] In this application embodiment, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent the existence of A alone, the simultaneous existence of A and B, or the existence of B alone. A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of singular or plural items. For example, at least one of a, b, and c can represent: the existence of a alone, the existence of b alone, the existence of c alone, the simultaneous existence of a and b, the simultaneous existence of a and c, the simultaneous existence of b and c, or the simultaneous existence of a, b, and c, where a, b, and c can be single or multiple.

[0055] In the description of this application, unless otherwise expressly defined, terms such as "setup," "installation," and "connection" should be interpreted broadly, and those skilled in the art can reasonably determine the specific meaning of the above terms in this application in conjunction with the specific content of the technical solution.

[0056] The power quality waveform pattern recognition method for distributed generation systems provided in this application can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms; the software can be an application that implements the power quality waveform pattern recognition method for distributed generation systems, but is not limited to the above forms.

[0057] This application can be applied to numerous general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via communication networks. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices. It should be noted that in various specific embodiments of this invention, when processing is required based on data related to the characteristics of an object (e.g., user attributes or sets of attribute information), permission or consent from the corresponding object is obtained first, and the collection, use, and processing of this data comply with relevant laws and standards. Furthermore, when the embodiments of the present invention need to obtain the attribute information of an object, they will obtain the separate permission or separate consent of the corresponding object through pop-up windows or redirection to a confirmation page. After obtaining the separate permission or separate consent of the corresponding object, they will then obtain the relevant data of the object necessary for the embodiments of the present invention to operate normally.

[0058] See Figure 1 , Figure 1This is a flowchart illustrating a power quality waveform pattern recognition method for a distributed generation system according to an embodiment of this application. The power quality waveform pattern recognition method for a distributed generation system provided in this embodiment includes, but is not limited to, steps S110 to S150, which are described below.

[0059] Step S110: Obtain the current waveform data of voltage and current in the power grid and the operating status of the power grid, and establish and update the background waveform characteristic reference based on the operating status of the power grid and the current waveform data.

[0060] Step S120: Based on the current waveform data and the background waveform feature benchmark, detect the deviation in the current waveform data to identify one or more potential events of interest, wherein the potential events of interest represent events in which the current waveform data deviates;

[0061] Step S130: In response to a potential event of concern, backtrack the high-resolution waveform data of the time period in which the potential event of concern occurred;

[0062] Step S140: Extract micro-features from the high-resolution waveform data to obtain micro-features, whereby the micro-features are used to characterize the details of potential events of interest.

[0063] Step S150: Combine the operating status to make attribution judgments on micro-features and obtain judgment results. The judgment results include distinguishing between benign transients, changes caused by component aging, or early signs of failure.

[0064] It should be noted that current waveform data refers to the voltage and current waveform information collected from the power grid in real time at a specific point in time or within a specific time period. This data is the foundation for power quality analysis. The operating state of the power grid refers to the operating condition of the power grid at a specific moment, such as normal operation, microgrid islanding switching, load changes, generator start-up and shutdown, etc. These states have a significant impact on waveform data. The background waveform characteristic benchmark is a reference model of voltage and current waveform characteristics established through long-term monitoring and statistical analysis of the power grid under normal operating conditions. It reflects the normal behavior pattern of the power grid and is dynamically updated according to the actual operating conditions of the power grid. Potential events of concern refer to events where there is a significant deviation between the current waveform data and the background waveform characteristic benchmark. These deviations may indicate power quality problems or system anomalies. High-resolution waveform data refers to waveform data collected at a higher sampling rate or with finer granularity during the period of a potential event of concern, used for in-depth analysis of event details. Micro-features are quantitative indicators extracted from high-resolution waveform data that can characterize the specific details of potential events of concern, such as transient energy, waveform distortion, and frequency variation. Attribution judgment refers to analyzing micro-features in conjunction with the operating status of the power grid to determine the root cause of potential events of concern and classify them as benign transients, changes caused by component aging, or early signs of failure.

[0065] In one embodiment, various methods can be employed to acquire current waveform data of voltage and current in the power grid, as well as the operating status of the power grid, and to establish and update a background waveform characteristic benchmark based on the operating status of the power grid and the current waveform data. Voltage and current data of the power grid can be collected periodically, and power grid operating status information can be obtained in conjunction with a SCADA system or sensor network. When establishing the background waveform characteristic benchmark, a moving average method can be used. That is, within each preset time window, statistical analysis is performed on the collected waveform data to calculate its average value, standard deviation, harmonic content, and other characteristics, and these characteristics are used as the background waveform characteristic benchmark. Over time, new waveform data is continuously added, and old data is gradually removed, thereby achieving dynamic updating of the background waveform characteristic benchmark. Statistical methods can be used to detect deviations in the current waveform data based on the current waveform data and the background waveform characteristic benchmark to identify one or more potential events of concern. The Euclidean distance or Mahalanobis distance between the current waveform data and the background waveform characteristic benchmark can be calculated. When this distance exceeds a preset deviation threshold, a deviation is considered to exist and it is identified as a potential event of concern. Deviation thresholds can be statistically analyzed based on historical data and set as the mean of the background waveform characteristic benchmark plus or minus three standard deviations. Alternatively, a control chart-based approach can be used, establishing a Shewhart or CUSUM control chart. A characteristic value of the current waveform data (such as RMS value or harmonic distortion rate) is compared with the background waveform characteristic benchmark; when a point on the control chart exceeds the control limit, it is identified as a potential event of interest. In responding to potential events of interest and tracing back to high-resolution waveform data during the period in which the event occurred, a data storage and retrieval system can be utilized. After the detection module identifies a potential event of interest, the system records the timestamp of the event. Subsequently, using this timestamp, the system can retrieve high-resolution waveform data from a high-speed data storage device for a period before and after the event. For micro-feature extraction from high-resolution waveform data, various signal processing techniques can be employed. For example, Fourier transforms can be performed on the high-resolution waveform data to analyze its spectral characteristics and extract harmonic content, interharmonic content, frequency shift, etc., as micro-features. To determine the cause of micro-features by combining them with operational status, rule-based expert systems or machine learning classifiers can be used. A series of pre-defined rules can be used to associate specific combinations of micro-features with specific operational states for attribution. For example, if micro-features show a voltage drop followed by rapid recovery, and the operational state is microgrid islanding switching, it is judged as a benign transient. If micro-features show a slow increase in harmonic content, and the operational state is within the normal aging trajectory of components, it is judged as a change caused by component aging. If micro-features show intermittent, low-amplitude abnormal fluctuations, and the operational state is outside the normal aging trajectory and not within microgrid islanding switching, it is judged as an early sign of a fault.In addition, machine learning models such as support vector machines (SVM), decision trees, or neural networks can be used to learn the mapping relationship between micro-features and attribution results by training on a large amount of historical data, thereby achieving automated attribution judgment.

[0066] It should be noted that this application acquires the real-time operating status and current waveform data of the power grid, and dynamically adjusts the background waveform characteristic benchmark accordingly, so that the definition of the normal waveform pattern can continuously adapt to the actual operating conditions of the power grid. When a deviation is detected between the current waveform data and the dynamically updated background waveform characteristic benchmark, the system can identify potential events of concern. Subsequently, by backtracking high-resolution waveform data and extracting micro-features, the system can capture the fine details of the event. More importantly, this application combines the operating status of the power grid to perform attribution judgment on micro-features, which can accurately distinguish between benign transients, changes caused by component aging, or early signs of faults. For example, during microgrid islanding switching, the system can identify the resulting voltage drop and rapid recovery of benign transients, avoiding false alarms. Under the normal aging trajectory of components, the system can identify new normal changes such as the slow increase in harmonic content caused by component aging, thereby making accurate attributions. For intermittent, low-amplitude precursory events that foreshadow initial faults, this application, due to its understanding of the power grid operating context and dynamic adaptability, can reliably capture and classify them, thereby providing early warnings and significantly improving the system's ability to provide predictive maintenance insights and proactive fault mitigation.

[0067] In this regard, this application further proposes the following steps for establishing and updating the background waveform characteristic reference based on the operating status of the power grid and the current waveform data:

[0068] When establishing and updating the background waveform characteristic benchmark based on the current waveform data of grid voltage and current based on the preset monitoring cycle, the current waveform data within each preset monitoring cycle is detected to obtain high-frequency transient energy and waveform distortion measurement.

[0069] When a high-frequency transient energy exceeds a preset threshold, or when waveform distortion measurement fluctuates within a preset time, the data segment within the preset monitoring period is marked as a potential abnormal data segment.

[0070] When the micro-features of a potential anomalous data segment deviate from the background waveform feature benchmark, and the deviation does not conform to the known benign transient or component aging-induced changes, the potential anomalous data segment is isolated, the isolated potential anomalous data segment is removed from the update calculation of the background waveform feature benchmark, and the background waveform feature benchmark is established and updated.

[0071] Specifically, in establishing and updating the background waveform characteristic benchmark, the current waveform data of the power grid voltage and current are first continuously monitored based on a preset monitoring period. This preset monitoring period can be set according to actual needs, for example, it can be several seconds, several minutes, or longer. Within each preset monitoring period, the current waveform data is analyzed to obtain high-frequency transient energy and waveform distortion metrics. High-frequency transient energy reflects whether there are rapidly changing transient components in the waveform, for example, by extracting high-frequency components and calculating their energy through methods such as wavelet transform and Fourier transform. Waveform distortion metrics are used to quantify the degree to which the waveform deviates from a standard sine wave, for example, by using indicators such as total harmonic distortion (THD) or specific harmonic content. Furthermore, the system continuously monitors these metrics. When the detected high-frequency transient energy exceeds a preset threshold, it indicates that a transient event may exist; or when the waveform distortion metric fluctuates significantly within a preset time, it also indicates that the waveform may be abnormal. Once one of these conditions is met, the data segment within that preset monitoring period is marked as a potentially abnormal data segment. The preset threshold and preset time can be set according to the characteristics of the power grid and experience. For flagged potentially anomalous data segments, their micro-features are further compared with the current background waveform feature benchmark to determine if any deviation exists. Micro-features are used to more finely characterize waveform details, and may include transient wavelet coefficients, waveform geometry (such as curvature), or energy spectral density. If the detected deviation does not conform to known benign transients (e.g., voltage drops caused by normal load switching) or changes caused by component aging (e.g., the slow decrease in output power of photovoltaic modules during normal aging), then the potentially anomalous data segment is considered to indeed contain an unexpected anomaly. In this case, the potentially anomalous data segment will be isolated and removed from the update calculation of the background waveform feature benchmark. In this way, the establishment and updating process of the background waveform feature benchmark can be ensured to be unaffected by short-term, atypical anomalies, thereby maintaining the purity and accuracy of the benchmark.

[0072] It should be noted that the solution proposed in this application effectively solves the problem of the benchmark being susceptible to transient events or short-term anomalies in traditional methods by introducing a mechanism for detecting, marking, judging, and isolating abnormal data segments during the establishment and updating of the background waveform characteristic benchmark. Specifically, by detecting high-frequency transient energy and waveform distortion measurements of the current waveform data within each preset monitoring period, waveform data segments that may contain anomalies can be preliminarily identified. Furthermore, by comparing the micro-features of these potential abnormal data segments with the background waveform characteristic benchmark and excluding changes caused by known benign transients or component aging, it is ensured that only truly atypical anomalies are isolated. Thus, when updating the background waveform characteristic benchmark, these isolated abnormal data segments can be removed from the calculation, thereby avoiding interference from abnormal data to the benchmark, and enabling the established and updated background waveform characteristic benchmark to more accurately reflect the normal operating mode of the power grid. Through the above technical solution, this application can significantly improve the accuracy and robustness of power quality waveform pattern recognition in distributed generation systems. Because the background waveform characteristic benchmark excludes interference from transient events and short-term anomalies during its establishment and updating process, it can more realistically and stably reflect the normal operating status of the power grid. This not only avoids the risk of misjudging normal transient events as faults, but also enables more sensitive detection of early fault precursors or component anomalies that deviate slightly from the benchmark. This provides a more reliable basis for decision-making in the operation and maintenance of distributed generation systems, effectively improving the efficiency of power quality monitoring and fault diagnosis.

[0073] In one embodiment, suppose that in a distributed generation system, the system is continuously establishing and updating a background waveform characteristic benchmark for the grid voltage waveform. During a preset monitoring period, a sudden connection of a large inductive load nearby causes a brief voltage drop in the grid, accompanied by high-frequency oscillations. At this time, the system first detects the voltage waveform data within this preset monitoring period. Calculations reveal that the high-frequency transient energy of this data segment significantly exceeds a preset threshold, and waveform distortion also fluctuates within a short period. Therefore, this data segment is marked as a potentially anomalous data segment. Next, the system extracts the micro-features of this potentially anomalous data segment, such as analyzing its transient wavelet coefficients, and compares them with the current background waveform characteristic benchmark. Through comparison, the system identifies that the deviation pattern matches the characteristics of a known benign transient event—a voltage drop caused by a large load switching. In this case, despite the deviation, because it conforms to a known benign transient pattern, the data segment is not isolated but is allowed to participate in the update calculation of the background waveform characteristic benchmark to ensure that the benchmark can adapt to changes in the normal operation of the grid. However, if the micro-features of a potentially anomalous data segment exhibit a persistent high-frequency noise that does not conform to any known benign transient or component aging trajectory, and its deviation exceeds the normal range, then the data segment will be isolated and removed from the update calculation of the background waveform characteristic benchmark. For example, if a persistent, non-periodic high-frequency harmonic component is detected, and it does not belong to normal load or grid background noise, then the data segment will be considered anomalous and not used for benchmark updates. In this way, the purity of the background waveform characteristic benchmark can be ensured, avoiding the inclusion of occasional or atypical anomalies in the benchmark, thereby improving the accuracy of subsequent anomaly detection.

[0074] In this regard, this application further proposes the following steps for establishing and updating the background waveform characteristic benchmark:

[0075] Obtain historical waveform reference values ​​of voltage and current in the power grid during historical periods;

[0076] The preset smoothing factor is adjusted according to the operating status of the power grid to obtain the adjusted preset smoothing factor;

[0077] The background waveform feature benchmark is established and updated by calculating the current waveform data, historical waveform benchmark values, and adjusted preset smoothing factors based on the exponentially weighted moving average method.

[0078] Specifically, acquiring historical waveform benchmark values ​​for grid voltage and current over a historical period means that the system stores and utilizes stable and normal waveform data of grid voltage and current over a past period as a reference. These historical waveform benchmark values ​​reflect the typical characteristics of the grid under different operating conditions. Their purpose is to provide a long-term, stable reference point for updating the background waveform characteristic benchmark, avoiding excessive influence of short-term fluctuations on the benchmark. Adjusting the preset smoothing factor according to the grid's operating state can be understood as dynamically adjusting the weighting parameters used for smoothing data and calculating moving averages based on the current operating mode of the grid (e.g., peak load, off-peak load, changes in distributed power generation, islanded operation, etc.). The preset smoothing factor determines the degree of influence of the current waveform data on the updating of the background waveform characteristic benchmark. A larger smoothing factor means that the current data has a greater impact on the benchmark, and the benchmark responds faster; a smaller smoothing factor means that the benchmark is more stable and less sensitive to short-term fluctuations. The aim is to enable the background waveform characteristic benchmark to adapt more flexibly to changes in the grid's operating state, remaining stable in a stable state and responding quickly in a changing state. In practical applications, the background waveform characteristic benchmark is established and updated by calculating the current waveform data, historical waveform benchmark values, and adjusted preset smoothing factors using the exponentially weighted moving average method. This involves employing a weighted averaging method where newer data is assigned greater weight, while older data has progressively less weight. By combining the current waveform data, historical waveform benchmark values, and smoothing factors adjusted according to the operating status, a dynamically updated background waveform characteristic benchmark can be calculated.

[0079] It should be noted that the solution proposed in this application solves the problems of slow response or oversensitivity that may exist in traditional methods when establishing background waveform characteristic benchmarks by introducing historical waveform benchmark values ​​as long-term references and dynamically adjusting the smoothing factor according to the power grid operating status, combined with the exponentially weighted moving average method for calculation. Specifically, the introduction of historical waveform benchmark values ​​provides a stable anchor point for benchmark updates, preventing the benchmark from drifting significantly during short-term drastic fluctuations. Simultaneously, adjusting the preset smoothing factor according to the power grid operating status allows the benchmark to intelligently adapt to different power grid environments: when the power grid is operating smoothly, a smaller smoothing factor can be used to make the benchmark more stable and reduce false alarms; when the power grid operating status changes frequently or requires rapid response, a larger smoothing factor can be used to enable the benchmark to capture new normal states more quickly. The exponentially weighted moving average method effectively balances the influence of current and historical data, ensuring the smoothness of the benchmark and its sensitivity to the latest trends. Through the above technical solutions, the established background waveform characteristic benchmark can more accurately and robustly reflect the actual operating status of the power grid. This scheme effectively avoids baseline distortion caused by changes in grid operating conditions, thereby significantly improving the accuracy of identifying potential events of concern and reducing false alarms and missed alarms. Furthermore, by dynamically adjusting the smoothing factor, the system can better adapt to the complexity and variability of distributed generation system grids, providing a more reliable foundation for subsequent micro-feature extraction and attribution judgment.

[0080] In one embodiment, assuming a distributed generation system grid operates more actively during the day due to high photovoltaic (PV) power generation, potentially leading to greater fluctuations; while at night, PV power generation is zero, and the grid is primarily powered by conventional sources, resulting in a relatively stable operation. When the system detects the grid operating actively during the day, it can adjust a preset smoothing factor to a relatively large value based on the grid's operating status (high distributed generation output, high load fluctuations). In this case, the exponentially weighted moving average method assigns greater weight to the current waveform data when calculating the background waveform characteristic benchmark, allowing the benchmark to adapt more quickly to the new normal caused by changes in PV output or load fluctuations, thus avoiding misidentifying normal fluctuations as anomalies. Conversely, when the system detects the grid operating stably at night, it can adjust the preset smoothing factor to a relatively small value based on the grid's operating status (low distributed generation output, low load fluctuations). In this case, the exponentially weighted moving average method assigns greater weight to historical waveform benchmark values, making the benchmark more stable and less sensitive to minor, random noise that may occur at night, effectively avoiding misidentifying normal small fluctuations as potential events of concern. By dynamically adjusting the smoothing factor and combining historical benchmark values ​​with the exponentially weighted moving average method, the background waveform feature benchmark can intelligently adapt to the characteristics of the power grid under different operating conditions, thereby improving the accuracy and robustness of the entire power quality waveform pattern recognition method.

[0081] In response, this application further proposes the above-mentioned adjustment of the preset smoothing factor according to the operating status of the power grid, resulting in an adjusted preset smoothing factor, including:

[0082] When the fluctuation amplitude of the current waveform data is detected to be less than or equal to the first preset threshold, the preset smoothing factor is increased to obtain the adjusted preset smoothing factor.

[0083] When the fluctuation amplitude of the current waveform data is detected to be greater than or equal to the second preset threshold, the preset smoothing factor is reduced to obtain the adjusted preset smoothing factor, wherein the first preset threshold is less than the second preset threshold.

[0084] Specifically, the preset smoothing factor is a key parameter in the exponentially weighted moving average method, and its value directly affects the update speed of the background waveform characteristic benchmark and its responsiveness to new data. The first and second preset thresholds are reference values ​​used to determine the fluctuation amplitude of the current waveform data; they are preset to distinguish different fluctuation states. Specifically, the first preset threshold defines a stable state with small waveform fluctuations, while the second preset threshold defines an unstable state with large waveform fluctuations. When the monitored fluctuation amplitude of the current waveform data is small, i.e., less than or equal to the first preset threshold, it indicates that the power grid is relatively stable. In this case, increasing the preset smoothing factor can make the update of the background waveform characteristic benchmark smoother, reducing over-response to small fluctuations and thus improving the benchmark's stability. Conversely, when the monitored fluctuation amplitude of the current waveform data is large, i.e., greater than or equal to the second preset threshold, it indicates that the power grid may be in a state of change or disturbance. In this case, decreasing the preset smoothing factor can make the background waveform characteristic benchmark adapt to new waveform changes more quickly, improving its responsiveness to changes in the power grid state. It is important to note that the first preset threshold is set to be less than the second preset threshold to ensure that the adjustment logic of the smoothing factor is clear and conflict-free within different fluctuation amplitude ranges.

[0085] It should be noted that the scheme in this application achieves adaptive updating of the background waveform characteristic benchmark by dynamically adjusting a preset smoothing factor. When the power grid is operating stably and the waveform fluctuation amplitude is small, increasing the smoothing factor makes the benchmark update process smoother, effectively suppressing the interference of noise and minor fluctuations on the benchmark, and ensuring the accuracy and stability of the benchmark. However, when the power grid operating state changes significantly and the waveform fluctuation amplitude is large, decreasing the smoothing factor allows the benchmark to respond to these changes more quickly, thereby avoiding misjudgments or omissions caused by benchmark update lag. This adaptive adjustment mechanism enables the background waveform characteristic benchmark to better reflect the actual operating conditions of the power grid, providing a more reliable reference for subsequent deviation detection and event identification.

[0086] In some embodiments described above in this application, a scheme is proposed for detecting deviations in current waveform data based on a background waveform feature benchmark, in order to identify one or more potential events of interest. Specifically, the step of detecting deviations in current waveform data based on a background waveform feature benchmark, in order to identify one or more potential events of interest, may include the following specific operations:

[0087] Determine the deviation threshold based on the background waveform characteristics;

[0088] Establish a statistical control chart for detection deviations;

[0089] Obtain the instantaneous value of the current waveform based on the current waveform data;

[0090] The preset instantaneous value of the waveform is obtained based on the background waveform feature benchmark;

[0091] Input the current waveform instantaneous value, the preset waveform instantaneous value, and the deviation threshold into the statistical control chart to detect deviations in the current waveform data in order to identify one or more potential events of concern.

[0092] Specifically, when detecting deviations in the current waveform data, a deviation threshold must first be determined based on the background waveform characteristic benchmark. This deviation threshold defines what degree of waveform change is considered a significant deviation. It can be set based on statistical methods by calculating the mean and standard deviation of the background waveform characteristic benchmark to ensure effective differentiation between normal fluctuations and abnormal events. To systematically detect deviations, a statistical control chart for deviation detection is established. A statistical control chart is a commonly used quality control tool that can visually display the trend of data points over time and indicate the upper and lower control limits, thereby assisting in determining whether the current waveform data is within the normal range. During deviation detection, the instantaneous value of the current waveform needs to be obtained based on the current waveform data, and a preset instantaneous value of the waveform needs to be obtained based on the background waveform characteristic benchmark. The instantaneous value of the current waveform refers to the value of the voltage or current waveform collected in real time from the power grid at a specific sampling moment, while the preset instantaneous value of the waveform refers to the waveform value predicted or expected based on the established background waveform characteristic benchmark at the same sampling moment. Subsequently, the instantaneous value of the current waveform, the preset instantaneous value of the waveform, and the deviation threshold are input into the statistical control chart. Statistical control charts calculate and plot corresponding statistics based on these input data, and compare them with preset control limits (determined by deviation thresholds). When a statistic exceeds the control limit, it indicates a significant deviation in the current waveform data, thereby identifying one or more potential events of concern.

[0093] It should be noted that the solution presented in this application provides a systematic and quantitative method for detecting deviations in power quality waveforms by introducing deviation thresholds and statistical control charts, and by comparing the current instantaneous waveform value with a preset instantaneous waveform value. This method can effectively compare the real-time data of the current waveform with long-term stable background characteristics, thereby accurately capturing any changes that exceed the normal fluctuation range. The use of statistical control charts makes the deviation detection process more visual and standardized, helping operators to quickly identify abnormal situations.

[0094] Specifically, the above-mentioned response to a potential event of concern, which involves tracing back high-resolution waveform data during the period in which the potential event of concern occurred, may include the following steps:

[0095] In response to potential events of concern, determine the timestamp of the event's occurrence;

[0096] Based on the timestamp, high-resolution waveform data of the period in which the potentially concerning event occurred can be traced back.

[0097] Specifically, when the system detects a deviation between the current waveform data and the background waveform feature benchmark, and identifies one or more potential events of interest, it needs to accurately determine the specific time point of the event, i.e., the occurrence timestamp. This timestamp can be accurate to milliseconds or even microseconds, uniquely identifying the start or critical moment of the event. The purpose of determining the occurrence timestamp is to accurately locate and extract detailed waveform data associated with the event. Specifically, tracing back to high-resolution waveform data during the period in which the potential event of interest occurred means that after determining the occurrence timestamp, the system retrieves and extracts the raw, uncompressed, or unsampled waveform data of the grid voltage and current for a period before and after the event from historical data storage based on that timestamp. High-resolution waveform data typically means a higher sampling rate and finer temporal granularity, capable of capturing minute changes and transient features in the waveform; these details are crucial for accurately analyzing the nature of the event. Through the above technical solution, it is possible to ensure that after identifying a potential event of interest, detailed waveform information for the period in which the event occurred can be obtained quickly and accurately. This high-resolution data backtracking capability greatly improves the accuracy of capturing event details, enabling subsequent micro-feature extraction to be more comprehensive and in-depth. This provides more reliable and refined data support for distinguishing between benign transients, changes caused by component aging, or early signs of failure, effectively avoiding misjudgments or omissions caused by insufficient data resolution.

[0098] Specifically, the above-mentioned micro-feature extraction from high-resolution waveform data yields micro-features, which are used to characterize the details of potential events of interest and can include the following methods:

[0099] Transient wavelet analysis is performed on high-resolution waveform data to obtain energy and amplitude in different frequency ranges to obtain micro-features;

[0100] Alternatively, waveform geometric property analysis can be performed on high-resolution waveform data to obtain the curvature of each sampling point in the high-resolution waveform data in order to obtain micro-features;

[0101] Alternatively, energy spectral density analysis can be performed on high-resolution waveform data to obtain the energy spectral density of the waveform and thus obtain micro-features.

[0102] Transient wavelet analysis is a time-frequency analysis method that aims to capture the transient characteristics of waveforms simultaneously in both the time and frequency domains by decomposing high-resolution waveform data into different frequency components. Wavelet transform yields energy and amplitude information within different frequency ranges, effectively revealing drastic changes in the waveform over short periods, such as voltage dips, swells, or transient harmonic distortions. These energy and amplitude values ​​can serve as micro-features to finely characterize the details of potential events of interest. Furthermore, waveform geometric analysis extracts micro-features by calculating the curvature of each sampling point in the high-resolution waveform data. Curvature is a geometric quantity describing the degree of curvature of a curve; for waveform data, changes in curvature can sensitively reflect abrupt changes in the local shape of the waveform. Additionally, energy spectral density analysis is a frequency domain analysis method that aims to reveal the energy distribution of a waveform at different frequencies. By performing Fourier transform or other spectral analysis methods on the high-resolution waveform data, the energy spectral density of the waveform can be obtained. Energy spectral density visually displays the intensity of harmonics, interharmonics, or other non-fundamental frequency components contained in the waveform, which are often important indicators of power quality problems. For example, an abnormal increase in energy at a specific frequency may indicate a certain fault mode. Using the energy spectral density of a waveform as a micro-feature helps to analyze the nature of potential events of interest from the perspective of frequency components.

[0103] In response, this application further proposes the following steps for attributing micro-features to their causes and obtaining the judgment results, based on the aforementioned operational status:

[0104] When the operating state is microgrid islanding switching and the micro-feature characterizes the current waveform data as having a drop that recovers within a preset time, and the drop amplitude is less than the preset amplitude, the judgment result is a benign transient.

[0105] When the operating status is a normal aging trajectory and the micro-features indicate an increase in the current waveform data, the judgment result is that the component aging is in a new normal.

[0106] When the operating status is an abnormal aging trajectory and not a microgrid islanding switch, and the deviation of the current waveform data in the micro-feature characterization exceeds the preset range, the judgment result is that it is an early sign of an initial fault.

[0107] Specifically, operational status refers to the operating condition of a distributed generation system at a specific point in time, such as microgrid islanding switching, grid-connected operation, normal operation, or maintenance. These operational statuses can be acquired and determined through information such as system controller signals, switch states, and power flow. Micro-features are quantitative indicators used to characterize the details of potential events of interest. They can be energy and amplitude obtained through transient wavelet analysis, curvature obtained through waveform geometric characteristic analysis, or energy spectral density obtained through energy spectral density analysis. These micro-features can meticulously reflect information such as the shape, duration, and frequency components of waveform deviations. Attribution judgment refers to classifying detected waveform deviation events into specific cause types according to preset rules or models. The judgment results aim to clearly distinguish the nature of the event, including benign transients, changes caused by component aging, or early signs of faults. Benign transients typically refer to short-lived fluctuations that are harmless to system operation, such as voltage drops or rises caused by system topology changes during microgrid islanding switching, but their amplitude is small and recovers within a short time. The new normal of component aging refers to the slow performance degradation caused by the long-term operation of components (such as inverters and photovoltaic panels) in distributed generation systems. This may manifest as a slow drift in waveform parameters or a gradual increase in specific harmonic content. Early signs of impending system failure, on the other hand, are early indications of impending system failure, with waveform deviations typically exhibiting abnormal patterns or fluctuations exceeding normal ranges. Specifically, when the system is in a microgrid islanding switching state, if the micro-features show a brief dip in the current waveform data that recovers within a preset time, and the amplitude of the dip is less than a preset amplitude, this event is considered a benign transient. This dip is a normal phenomenon during islanding switching and should not be misjudged as a fault. The preset time can be set according to the system response speed and switching characteristics, for example, from tens to hundreds of milliseconds. The preset amplitude can be set according to the system voltage level and the allowable transient fluctuation range, for example, 5% to 10% of the nominal voltage. Furthermore, when the system operating state is identified as a normal aging trajectory, if the micro-features indicate a continuous, slow increasing trend in the current waveform data, this is usually considered a new normal of component aging. Normal aging trajectories can be determined by long-term monitoring of component performance parameters (such as efficiency and output power). Increases in waveform data may manifest as a slow rise in harmonic distortion, a gradual increase in specific components of voltage or current, reflecting the natural degradation process of component performance. Furthermore, when the operating state is neither a normal aging trajectory nor a microgrid islanding switchover, and the deviation of the current waveform data from the micro-features exceeds a preset range, this event will be judged as an early warning sign of a fault. In this case, the waveform deviation pattern does not conform to known benign transient or aging characteristics, and its degree has reached a level that may indicate a fault.The preset range can be determined based on historical data, system design specifications, and fault mode analysis. For example, when the harmonic distortion rate suddenly increases significantly, or when there are continuous abnormal fluctuations in voltage / current, and these fluctuations do not conform to the characteristics of any known normal operating conditions, they can be regarded as early signs of faults.

[0108] In response, this application further proposes that after attributing micro-features to their operational status and obtaining the judgment result, the following steps are included:

[0109] The judgment results will generate an alarm report, which includes the event time, location, anomaly type, and relevant micro-feature values.

[0110] An alarm report, including the event time, location, anomaly type, and relevant micro-feature values, will be sent to the user.

[0111] Specifically, an alarm report is a structured information carrier designed to present the results of power quality waveform pattern recognition in a clear and standardized format. Key information elements included in the alarm report include event time, location, anomaly type, and relevant micro-feature values. Event time accurately records the specific moment a potentially concerning event occurs, crucial for subsequent problem tracing and analysis. Location indicates the specific physical location or device node where the power quality anomaly occurred, such as an inverter, energy storage unit, or grid connection point in a distributed generation system, facilitating rapid identification of the problem's source. Anomaly type clarifies the classification of the judgment result, such as whether it is a benign transient, a change caused by component aging, or an early sign of a fault, providing users with a preliminary diagnostic conclusion. Relevant micro-feature values ​​provide detailed technical evidence supporting the judgment result, such as high-frequency transient energy, waveform distortion measurement, energy and amplitude from transient wavelet analysis, and curvature or energy spectral density from waveform geometric characteristic analysis. These micro-feature values ​​help professionals gain a deeper understanding of the nature and extent of the anomaly. The alarm report, including the above information, is sent to the user and transmitted to the designated recipients through various communication channels. Notifications can be sent via SMS, email, mobile application push notifications, SCADA system integration interfaces, or dedicated monitoring platform interfaces. The purpose is to ensure that relevant maintenance personnel, system administrators, or decision-makers receive timely notifications once potential power quality problems are identified, enabling them to respond quickly and take necessary inspection, maintenance, or intervention measures to prevent further deterioration of potential faults or adverse impacts on grid operation.

[0112] See Figure 2 , Figure 2 This is a schematic diagram of a distributed generation system power quality waveform pattern recognition system according to an embodiment of this application. The distributed generation system power quality waveform pattern recognition system 200 includes:

[0113] The acquisition module 210 is used to acquire the current waveform data of voltage and current in the power grid and the operating status of the power grid, and to establish and update the background waveform feature reference based on the operating status of the power grid and the current waveform data.

[0114] The detection module 220 is used to detect deviations in the current waveform data based on the current waveform data and the background waveform feature benchmark, so as to identify one or more potential events of interest, wherein the potential events of interest represent events in which the current waveform data deviates.

[0115] The backtracking module 230 is used to respond to a potential event of interest by backtracking high-resolution waveform data during the period in which the potential event of interest occurred.

[0116] The extraction module 240 is used to extract micro-features from high-resolution waveform data to obtain micro-features, wherein the micro-features are used to characterize the details of potential events of interest.

[0117] The judgment module 250 is used to combine the operating status to perform attribution judgment on micro-features and obtain judgment results. The judgment results include distinguishing between benign transients, changes caused by component aging, or early signs of failure.

[0118] It should be noted that the information interaction and execution process between the above modules are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, which will not be repeated here.

[0119] It will be understood by those skilled in the art that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, as is known to those skilled in the art, communication media typically include computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0120] The foregoing has provided a detailed description of the preferred embodiments of this application. However, this application is not limited to the above-described embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this application. All such equivalent modifications or substitutions are included within the scope defined in this application.

Claims

1. A method for identifying power quality waveform patterns in a distributed generation system, characterized in that, include: Acquire the current waveform data of voltage and current in the power grid and the operating status of the power grid, and establish and update the background waveform feature benchmark based on the operating status of the power grid and the current waveform data; Based on the current waveform data and the background waveform feature benchmark, deviations in the current waveform data are detected to identify one or more potential events of interest, wherein the potential events of interest characterize events in which deviations occur in the current waveform data; In response to the potential event of interest, high-resolution waveform data of the time period in which the potential event of interest occurred is retrieved. Micro-features are extracted from the high-resolution waveform data to obtain micro-features, wherein the micro-features are used to characterize the details of the potential event of interest; Based on the operating status, the micro-features are attributed and judged to obtain the judgment result, which includes distinguishing between benign transients, changes caused by component aging, or early signs of failure.

2. The method according to claim 1, characterized in that, The step of establishing and updating the background waveform feature benchmark based on the operating status of the power grid and the current waveform data includes: When establishing and updating the background waveform feature benchmark based on the current waveform data of the power grid voltage and current based on the preset monitoring cycle, the current waveform data within each preset monitoring cycle is detected to obtain high-frequency transient energy and waveform distortion measurement. When the high-frequency transient energy is detected to exceed a preset threshold, or when the waveform distortion measurement fluctuates within a preset time, the data segment within the preset monitoring period is marked as a potential abnormal data segment. When the micro-features of the potential abnormal data segment deviate from the background waveform feature benchmark, and the deviation does not conform to the known benign transient or component aging-induced changes, the potential abnormal data segment is isolated, the isolated potential abnormal data segment is deleted from the update calculation of the background waveform feature benchmark, and the background waveform feature benchmark is established and updated.

3. The method according to claim 1, characterized in that, The step of establishing and updating the background waveform feature benchmark based on the operating status of the power grid and the current waveform data includes: Obtain historical waveform reference values ​​of voltage and current of the power grid during the historical period; The preset smoothing factor is adjusted according to the operating status of the power grid to obtain the adjusted preset smoothing factor; The background waveform feature benchmark is established and updated by calculating the current waveform data, the historical waveform benchmark value, and the adjusted preset smoothing factor based on the exponentially weighted moving average method.

4. The method according to claim 3, characterized in that, The step of adjusting the preset smoothing factor according to the operating state of the power grid to obtain the adjusted preset smoothing factor includes: When the fluctuation amplitude of the current waveform data is detected to be less than or equal to the first preset threshold, the preset smoothing factor is increased to obtain the adjusted preset smoothing factor. When the fluctuation amplitude of the current waveform data is detected to be greater than or equal to the second preset threshold, the preset smoothing factor is reduced to obtain the adjusted preset smoothing factor, wherein the first preset threshold is less than the second preset threshold.

5. The method according to claim 1, characterized in that, The step of detecting deviations in the current waveform data based on the current waveform data and the background waveform feature benchmark to identify one or more potential events of interest includes: Determine the deviation threshold based on the background waveform feature benchmark; Establish a statistical control chart for detection deviations; The instantaneous value of the current waveform is obtained based on the current waveform data; The preset instantaneous value of the waveform is obtained based on the background waveform feature reference; The current waveform instantaneous value, the preset waveform instantaneous value, and the deviation threshold are input into the statistical control chart to detect deviations in the current waveform data in order to identify one or more potential events of interest.

6. The method according to claim 1, characterized in that, The response to the potential event of interest involves backtracking high-resolution waveform data during the time period in which the potential event of interest occurred, including: In response to the potential event of concern, determine the timestamp of the occurrence of the potential event of concern; Based on the occurrence timestamp, high-resolution waveform data of the period in which the potential event of concern occurred is traced back.

7. The method according to claim 1, characterized in that, The high-resolution waveform data is subjected to micro-feature extraction to obtain micro-features, wherein the micro-features are used to characterize the details of the potential event of interest, including: Transient wavelet analysis is performed on the high-resolution waveform data to obtain the energy and amplitude in different frequency ranges to obtain micro-features; Alternatively, waveform geometric property analysis can be performed on the high-resolution waveform data to obtain the curvature of each sampling point in the high-resolution waveform data in order to obtain micro-features; Alternatively, energy spectral density analysis can be performed on the high-resolution waveform data to obtain the energy spectral density of the waveform and thus obtain micro-features.

8. The method according to claim 1, characterized in that, The attribution judgment of the micro-features based on the operating state, and the resulting judgment, include: When the operating state is microgrid islanding switching and the micro-feature indicates that the current waveform data has a drop and recovers within a preset time, and the magnitude of the drop is less than the preset magnitude, the judgment result is a benign transient. When the operating state is a normal aging trajectory and the micro-feature indicates that the current waveform data has increased, the judgment result is a new normal for component aging. When the operating state is an abnormal aging trajectory and not a microgrid islanding switch, and the micro-feature indicates that the deviation of the current waveform data exceeds the preset range, the judgment result is that it is an early sign of an initial fault.

9. The method according to claim 1, characterized in that, After combining the operating state to perform attribution judgment on the micro-features and obtain the judgment result, the process includes: The judgment result is used to generate an alarm report, wherein the alarm report includes the event time, location, anomaly type and relevant micro-feature values; An alarm report, including the event time, location, anomaly type, and relevant micro-feature values, will be sent to the user.

10. A power quality waveform pattern recognition system for a distributed generation system, characterized in that, include: The acquisition module is used to acquire the current waveform data of voltage and current in the power grid and the operating status of the power grid, and to establish and update the background waveform feature reference based on the operating status of the power grid and the current waveform data. The detection module is used to detect deviations in the current waveform data based on the current waveform data and the background waveform feature benchmark, so as to identify one or more potential events of interest, wherein the potential events of interest characterize events in which deviations occur in the current waveform data; The backtracking module is used to respond to the potential event of concern by backtracking the high-resolution waveform data of the time period in which the potential event of concern occurred; An extraction module is used to extract micro-features from the high-resolution waveform data to obtain micro-features, wherein the micro-features are used to characterize the details of the potential event of interest; The judgment module is used to perform attribution judgment on the micro-features in combination with the operating state and obtain the judgment result. The judgment result includes distinguishing between benign transients, changes caused by component aging, or early signs of failure.

Images