Method and device for controlling power quality of ac-dc distribution network

Abstract

The application discloses a kind of AC-DC distribution network power quality management and control method and equipment, comprising: collecting power quality data in AC-DC distribution network, and the characteristic analysis is carried out to power quality data, obtains current characteristic parameter and voltage abnormality mark; based on current characteristic parameter and voltage abnormality mark, power quality data is differentially compressed processing; compressed data and characteristic information are transmitted by redundant channel containing main, standby communication link; according to the data generated power quality alarm information transmitted, and according to this information, power grid operation decision control is carried out.The application collects the power quality data of a plurality of internet-of-things power quality sensors in AC-DC distribution network, carries out harmonic analysis to the current waveform of the power quality data, guarantees the accuracy and robustness of feature extraction, to meet the problem of modern AC-DC hybrid distribution network to high-precision, high-reliability and fast response power quality management and control demand.

Description

Technical Field

[0001] This invention relates to the field of quality control technology, and in particular to a method and equipment for controlling the power quality of AC / DC distribution networks. Background Technology

[0002] With the large-scale integration of distributed power sources, energy storage devices, and flexible interconnection devices, hybrid AC / DC distribution networks, while providing efficient and reliable power supply, also face increasingly severe power quality problems, such as harmonic pollution, voltage fluctuations, frequency deviations, and transient impacts. To ensure power quality, it is usually necessary to monitor and control the voltage, current, and power at each node of the distribution network in real time.

[0003] In existing technologies, online power quality monitoring terminals can collect voltage and current waveforms on the busbar to achieve local monitoring. However, due to the limited number of monitoring nodes, the data is mostly local measurements, making it difficult to accurately reflect the overall power quality status of the entire AC / DC distribution network, resulting in insufficient data coverage.

[0004] To improve monitoring coverage, some technical solutions introduce Internet of Things (IoT) sensor networks. These networks collect power grid operation data through a large number of low-cost sensor nodes and transmit it via wireless networks, achieving broader power quality sensing. However, IoT networks are generally based on low-power wide-area networks (LPWANs) and cellular networks, which suffer from limited bandwidth, latency fluctuations, and data packet loss. Particularly in the dynamic operation scenarios of AC / DC distribution networks, millisecond-level synchronization cannot be guaranteed, meaning that abnormal power quality voltage waveforms such as voltage drops and harmonic surges may not be captured in time, affecting rapid control and dynamic adjustment.

[0005] Therefore, existing power quality control technologies still have shortcomings in terms of monitoring coverage, real-time performance, and reliability, making it difficult to meet the high-precision, high-reliability, and fast-response power quality control requirements of modern AC / DC hybrid distribution networks. Summary of the Invention

[0006] This invention provides a method and equipment for power quality control in AC / DC distribution networks. Its main purpose is to solve the problem that existing power quality control technologies are still insufficient in terms of monitoring coverage, real-time performance and reliability, making it difficult to meet the requirements of modern AC / DC hybrid distribution networks for high-precision, high-reliability and fast-response power quality control.

[0007] In a first aspect, to achieve the above objectives, the present invention provides a power quality control method for AC / DC distribution networks, comprising:

[0008] The system collects power quality data from AC / DC distribution networks and performs feature analysis on the data to obtain current characteristic parameters and voltage anomaly indicators. Based on the current characteristic parameters and voltage anomaly indicators, it performs differentiated compression processing on the power quality data. The compressed data and feature information are transmitted through a redundant channel containing primary and backup communication links. Power quality alarm information is generated based on the transmitted data, and power grid operation decision control is performed based on this information.

[0009] The feature analysis disclosed herein includes: performing harmonic analysis on the current waveform to extract harmonic feature parameters; and performing adaptive threshold analysis on voltage fluctuations to generate voltage anomaly indicators.

[0010] The differentiated compression processing disclosed herein includes: judging the abnormal state of each data frame in the power quality data based on current characteristic parameters and voltage anomaly indicators; according to the judgment result, applying a lossy compression strategy to normal data frames and a lossless compression strategy to abnormal data frames, and merging and outputting the compressed data.

[0011] In this disclosure, the current waveform is subjected to harmonic analysis to extract harmonic characteristic parameters. Specifically, the power quality data is preprocessed to obtain a standardized current waveform signal; the standardized current waveform signal is segmented into frames to obtain an analysis frame signal; the analysis frame signal is subjected to spectral analysis to obtain the amplitude and phase parameters of the fundamental frequency and its harmonic frequencies; and the content rate of each harmonic and the total harmonic distortion rate are calculated based on the amplitude and phase parameters.

[0012] The adaptive threshold analysis of voltage fluctuations and the generation of voltage anomaly labels disclosed herein specifically involves: acquiring historical voltage waveforms and calculating the historical voltage mean and standard deviation; real-time acquisition of voltage waveforms and extraction of instantaneous voltage values; calculating the fluctuation amplitude based on the historical voltage mean and instantaneous voltage values; generating a dynamic fluctuation threshold range based on the historical voltage mean, historical voltage standard deviation, and a preset fluctuation factor; determining whether the fluctuation amplitude is within the dynamic fluctuation threshold range; if within the range, determining it as a normal voltage waveform; if outside the range, determining it as an abnormal voltage waveform and generating a voltage anomaly label.

[0013] The differential compression processing of power quality data disclosed herein includes: determining frame by frame whether a target frame in the power quality data is marked as voltage abnormality or harmonic abnormality, wherein the harmonic abnormality is obtained by comparing the amplitude of the current harmonic characteristics with a preset amplitude threshold; if the target frame has no voltage abnormality or harmonic abnormality, it is marked as a normal frame, and differentially encoded and lossy compressed to obtain a compressed normal waveform; if the target frame has any abnormality, it is marked as an abnormal frame, and losslessly compressed to obtain a compressed abnormal waveform; and all compressed normal waveforms and compressed abnormal waveforms are summarized to generate a compressed waveform.

[0014] The specific method for obtaining the compressed abnormal waveform disclosed herein is as follows: performing wavelet transform on the abnormal frame power quality data to obtain a multi-scale wavelet coefficient set; based on the energy distribution of the wavelet coefficient set, selecting a set of significant coefficients with significant energy; quantizing and losslessly encoding the set of significant coefficients to obtain a set of encoded coefficients; performing inverse transform on the encoded coefficients to generate a sparse approximate waveform, and calculating the residual signal between the original waveform and the sparse approximate waveform; performing lossless compression on the residual signal to obtain a compressed residual; and combining the encoded coefficients and the compressed residual as the compressed abnormal waveform.

[0015] The present disclosure describes the process of generating power quality alarm information based on transmitted data and making power grid operation decisions and controls based on this information, which includes: weighted fusion of current characteristic parameters, voltage anomaly identifiers, and compressed waveforms to generate a fused dataset; formatting and encapsulating the fused dataset and embedding a compressed fingerprint code generated based on the compressed waveform to obtain an updated dataset; generating an alarm data packet containing the updated dataset and its unique identifier; detecting the availability of the primary wireless link in the redundant communication channels and adaptively selecting the primary wireless link or the backup wired channel based on the detection results; and uploading the alarm data packet to the dispatch center through the selected channel.

[0016] The present disclosure further includes the following steps: generating power quality alarm information based on transmitted data and making power grid operation decision-making and control based on the information; calculating event risk control values ​​based on alarm data packets and determining the priority of abnormal events; matching candidate control schemes in a preset knowledge base to form an action space based on the priority; collecting real-time operating parameters of the distribution network to form a state space; constructing a reinforcement learning optimization model based on the action space, state space, and state transition probabilities; using the reinforcement learning optimization model to iterate the candidate control schemes; stopping the iteration when the accumulated reward reaches a preset threshold and outputting the optimal control scheme; converting the optimal control scheme into control commands and executing them to obtain power quality management and control results.

[0017] Secondly, the present invention also provides an electronic device, the electronic device comprising:

[0018] At least one processor; and,

[0019] A memory communicatively connected to the at least one processor; wherein,

[0020] The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the aforementioned AC / DC power quality control method for power distribution networks.

[0021] This invention discloses a power quality control method and device for AC / DC distribution networks. By collecting power quality data from several IoT power quality sensors in the AC / DC distribution network, and performing harmonic analysis on the current waveform of the power quality data, the accuracy and robustness of feature extraction are ensured, exhibiting good comparability and standardization. Alarm data packets are uploaded through the redundant communication channels. Combined with the intelligent selection mechanism of the redundant communication channels, this ensures reliable transmission of alarm data even in the event of main link failure or interference. This not only improves the accuracy and response speed of power quality anomaly alarms but also enhances the reliability, fault tolerance, and security of data transmission. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating a power quality control method for AC / DC distribution networks according to an embodiment of the present invention.

[0024] Figure 2 This is a schematic diagram of a power quality control system for AC / DC distribution networks provided in an embodiment of the present invention. Detailed Implementation

[0025] To enable those skilled in the art to better understand the technical solutions of this disclosure, and to fully understand and implement the process of how this disclosure applies technical means to solve technical problems and achieve corresponding technical effects, the technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, not all embodiments. The embodiments of this disclosure and the various features within them can be combined with each other without conflict, and the resulting technical solutions are all within the protection scope of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without creative effort should fall within the protection scope of this disclosure.

[0026] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this disclosure described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0027] This application provides a method for power quality control in AC / DC distribution networks. The executing entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the system provided in this application: a server, a terminal, etc. In other words, the method for power quality control in AC / DC distribution networks can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides 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 (CDNs), and big data and artificial intelligence platforms.

[0028] Reference Figure 1 The diagram shown is a flowchart illustrating a power quality control method for AC / DC distribution networks according to an embodiment of the present invention. In this embodiment, the power quality control method for AC / DC distribution networks includes:

[0029] S1. Collect power quality data in AC / DC distribution networks, perform feature analysis on the power quality data, and obtain current characteristic parameters and voltage anomaly indicators;

[0030] Specifically, the steps are as follows:

[0031] S11: Collect power quality data from several IoT power quality sensors in the AC / DC power distribution network, perform harmonic analysis on the current waveform of the power quality data, and obtain a set of current harmonic characteristics;

[0032] S12: Perform adaptive threshold determination on the voltage fluctuation of the power quality data, and filter out abnormal voltage waveforms based on the determination result to generate voltage abnormality labels.

[0033] In this embodiment of the invention, an AC / DC distribution network refers to a distribution network where both AC and DC power supplies exist simultaneously. For example, in an industrial park: AC side: industrial loads, distributed photovoltaic systems, flexible AC interconnection devices; DC side: data centers, charging piles, supercapacitor energy storage. IoT power quality sensors are distributed in the AC / DC distribution network at AC buses, DC buses, distributed power source access points, or key load nodes to collect power quality data. Power quality data are indicators describing the power supply quality of a power system, including but not limited to: voltage waveform (stability, fluctuations, sags / boosts), current waveform (harmonic content, inrush current), power factor, and frequency deviation.

[0034] In AC / DC hybrid distribution networks, IoT power quality sensors are installed on AC buses, DC buses, distributed power source access points, and critical load terminals. These sensors are electrically isolated and calibrated to provide node IDs and location information. The sampling rate and range are configured according to the operating scenario. Under normal conditions, current and voltage waveforms are sampled at 10kHz. When encountering disturbances, the sampling rate can be adaptively increased to a higher level. Anti-aliasing filters ensure data accuracy. The collected power quality data includes timestamps and node information, and waveforms of a certain duration are retained in a local cyclic buffer to support backtracking. This enables high-precision and continuous acquisition of power quality data in the distribution network.

[0035] In detail, the harmonic analysis of the current waveform of the power quality data to obtain a set of current harmonic characteristics includes:

[0036] The power quality data is timestamped to obtain aligned quality data;

[0037] The alignment quality data is subjected to noise removal using a preset adaptive filter to obtain denoised quality data.

[0038] The denoised quality data is normalized to obtain normalized quality data;

[0039] Extract the current waveform from the normalized quality data;

[0040] Obtain the waveform division length, window function, and overlap rate, and divide the current waveform into several initial frame signals according to the waveform division length;

[0041] The initial frame signal is optimized using the window function and the overlap rate to obtain the analysis frame signal;

[0042] Perform a Fast Fourier Transform on the analyzed frame signal to obtain the amplitude spectrum and phase spectrum for each frequency;

[0043] The fundamental frequency is determined from the amplitude spectrum and the phase spectrum, and the fundamental amplitude and fundamental phase of the fundamental frequency are identified.

[0044] Obtain the integer harmonics of the fundamental frequency, and extract the harmonic amplitude and phase of the integer harmonics.

[0045] The harmonic content of each integer octave is calculated based on the fundamental frequency amplitude and the octave amplitude.

[0046] Calculate the total harmonic distortion rate based on the harmonic content rate;

[0047] The fundamental amplitude, the fundamental phase, the octave amplitude, the octave phase, the harmonic content, and the total harmonic distortion rate are used to form a current harmonic characteristic set.

[0048] In detail, since the sampling time of IoT power quality sensors may be biased, the original power quality data may be out of sync. Using a unified clock source as a reference, the time stamps of the power quality data collected by each IoT power quality sensor are interpolated or compensated to ensure that the data at the same sampling time can correspond on the same time axis, thereby obtaining time-consistent aligned quality data.

[0049] Alignment quality data often contains power frequency interference, electromagnetic environmental noise, and sensor measurement errors. By using an adaptive filtering algorithm to adjust the filter weights in real time, the filtering parameters can be automatically optimized according to the statistical characteristics of the input signal. This allows the useful signal characteristics to be preserved to the maximum extent while suppressing random noise, ultimately resulting in denoised quality data.

[0050] Based on the overall distribution of the denoised quality data, obtain the statistical characteristics such as the maximum and minimum values, mean and standard deviation of the denoised quality data. Then, use methods such as min-max normalization and Z-score standardization to scale or standardize the denoised quality data so that the numerical range of the denoised quality data is mapped to a uniform interval or distribution, thereby obtaining normalized quality data.

[0051] By setting the sampling frequency and time series index, continuous power quality data is restored into a current signal sequence in time order, and the curve of the current amplitude changing with time is extracted from the current signal sequence to obtain the current waveform.

[0052] The continuous current waveform is sequentially divided into several initial frame signals based on the waveform division length. The edges of each initial frame signal are then weighted using a predefined window function to reduce spectral leakage caused by boundary effects. Suitable window functions include rectangular windows, Hanning windows, Hamming windows, and Blackman windows. Adjacent frames are then partially overlapped and stitched together based on the overlap rate, thereby improving the accuracy of frequency domain analysis while maintaining temporal continuity. The result is a smooth analysis frame signal suitable for subsequent feature extraction.

[0053] The Fast Fourier Transform algorithm is used to perform discrete frequency domain decomposition on the analysis frame signal, calculate the complex spectrum corresponding to each discrete frequency point, and extract the magnitude of the spectrum as the amplitude spectrum and the phase angle as the phase spectrum, thereby realizing a comprehensive characterization of the frequency domain characteristics of the current signal.

[0054] After acquiring the amplitude and phase spectra, the frequency component closest to the preset power frequency is searched and determined within the frequency domain as the fundamental frequency. Amplitude information is extracted from the spectral position corresponding to the fundamental frequency as the fundamental amplitude, and phase information is extracted as the fundamental phase. After determining the fundamental frequency, the positions of each octave point are calculated sequentially according to integer multiples of the fundamental frequency, and the corresponding frequency component information is extracted from the amplitude and phase spectra, thereby obtaining the octave point amplitude and phase.

[0055] The harmonic content is calculated by dividing the amplitude of each octave point by the fundamental frequency amplitude, as shown in the following formula:

[0056]

[0057] in, Indicates the first Harmonic content at integer multiples of frequencies Indicates the first The amplitude of the frequency at an integer multiple of octaves. This represents the fundamental frequency amplitude. Harmonic content is a key indicator of power quality, used to represent the ratio of harmonic energy to fundamental frequency energy at a given octave.

[0058] The formula for calculating the total harmonic distortion rate is shown below:

[0059]

[0060] in, Indicates the first Harmonic content at integer multiples of frequencies Represents the total number of integer multiples of frequency. This represents the total harmonic distortion (THD). The THD reflects the overall level of harmonic energy in the entire current waveform.

[0061] Through a complete time-domain preprocessing and frequency-domain analysis process, the fundamental and harmonic characteristics of the current signal can be accurately extracted after noise elimination and normalization of comparison differences. The power quality status is quantified in the form of harmonic content rate and total harmonic distortion rate, ensuring the accuracy and robustness of feature extraction. The resulting set of current harmonic features not only comprehensively covers key parameters such as amplitude and phase, but also has good comparability and standardization effect, thus providing a reliable basis for subsequent power quality monitoring, anomaly diagnosis and optimized control.

[0062] For S12: An adaptive threshold is applied to the voltage fluctuations in the power quality data, and voltage anomaly waveforms are selected based on the threshold result to generate voltage anomaly tags. Specifically:

[0063] In this embodiment of the invention, voltage fluctuation detection is performed on the collected power quality data. Voltage fluctuation limits under different operating conditions are dynamically calculated using an adaptive threshold algorithm. The actual voltage waveform is compared with the voltage fluctuation limits. If the voltage fluctuation limits are exceeded, the waveform is determined to be abnormal, and a corresponding voltage abnormality label is automatically generated, thereby achieving accurate identification and labeling of voltage abnormalities.

[0064] Specifically, the adaptive threshold determination of voltage fluctuations in the power quality data, and the filtering of abnormal voltage waveforms based on the determination results to generate abnormal voltage tags, includes:

[0065] Obtain the historical voltage waveform of historical power quality data, and calculate the historical voltage mean and historical voltage standard deviation based on the historical voltage waveform;

[0066] Extract the voltage waveform of the normalized quality data and obtain the instantaneous voltage value of the voltage waveform in each frame;

[0067] The fluctuation amplitude of the voltage waveform in each frame is calculated based on the historical average voltage value and the instantaneous voltage value;

[0068] A dynamic fluctuation threshold range is generated using the historical voltage mean, the historical voltage standard deviation, and a preset fluctuation factor.

[0069] Determine whether the fluctuation amplitude is greater than or equal to the lower limit of the dynamic fluctuation threshold range, and less than or equal to the upper limit of the dynamic fluctuation threshold range;

[0070] If the fluctuation amplitude is greater than or equal to the lower limit of the dynamic fluctuation threshold range and less than or equal to the upper limit of the dynamic fluctuation threshold range, then the voltage waveform corresponding to the fluctuation amplitude is determined to be a normal voltage waveform.

[0071] If the fluctuation amplitude is less than the lower limit of the dynamic fluctuation threshold range or greater than the upper limit of the dynamic fluctuation threshold range, the voltage waveform corresponding to the fluctuation amplitude is determined to be a voltage abnormal waveform, and a voltage abnormality label is generated based on the fluctuation amplitude.

[0072] In detail, historical voltage waveform sequences are extracted from historical power quality data, and the historical voltage waveform sequences are preprocessed to remove outliers and noise. Statistical analysis is performed on the preprocessed historical voltage waveforms in the time dimension to calculate the mean and standard deviation of the historical voltage waveform sequences, thereby obtaining reference indicators reflecting the historical voltage level and fluctuation characteristics.

[0073] The voltage waveform sequence is extracted from the normalized power quality data and divided into several analysis frames. The voltage waveform is processed point by point in each analysis frame to obtain the instantaneous voltage value at the corresponding time point.

[0074] Using the historical average voltage as a reference, the deviation of the instantaneous voltage value in each analysis frame is calculated point by point, and the deviation is used as the fluctuation amplitude of the voltage waveform of the corresponding analysis frame, thereby quantifying the fluctuation of the voltage waveform of each frame relative to the historical average voltage.

[0075] The formula for calculating the dynamic fluctuation threshold range using the historical voltage mean, historical voltage standard deviation, and preset fluctuation factor is shown below:

[0076]

[0077]

[0078] in, This indicates the upper limit of the dynamic fluctuation threshold range. Indicates volatility factor. Indicates the historical voltage standard deviation. This indicates the lower limit of the dynamic fluctuation threshold range.

[0079] The fluctuation amplitude of each frame of voltage waveform is compared with the upper and lower limits of the dynamic fluctuation threshold range. When the fluctuation amplitude is within the dynamic fluctuation threshold range, the corresponding voltage waveform is judged as normal. When the fluctuation amplitude is lower than the lower limit of the dynamic fluctuation threshold range or higher than the upper limit of the dynamic fluctuation threshold range, the corresponding voltage waveform is judged as abnormal. A corresponding voltage abnormality label is generated based on the fluctuation amplitude. The voltage abnormality label can be used to identify the location and degree of abnormal waveform.

[0080] By combining historical voltage statistics with real-time instantaneous voltage fluctuations, adaptive judgment of voltage anomalies is achieved. This does not rely on fixed thresholds, allowing the thresholds to be dynamically adjusted according to the grid's operating status. This improves the accuracy and sensitivity of voltage anomaly detection. The generated voltage anomaly tags can accurately identify the location and severity of abnormal waveforms, providing a reliable basis for power quality analysis, fault location, and subsequent regulation, and significantly enhancing the intelligence and reliability of power quality monitoring.

[0081] S3. Based on current characteristic parameters and voltage anomaly indicators, power quality data is subjected to differentiated compression processing. Specifically, power quality data is classified and compressed according to the current harmonic characteristic set and voltage anomaly label to obtain compressed waveforms.

[0082] In this embodiment of the invention, the current harmonic feature set and voltage anomaly label of power quality data are used as classification criteria to divide power quality data into normal frames and abnormal frames. Based on the classification results, an appropriate compression algorithm is applied to the waveform corresponding to each type of frame to generate compressed waveforms.

[0083] In detail, the step of classifying and compressing the power quality data according to the current harmonic feature set and the voltage anomaly label to obtain a compressed waveform includes:

[0084] When any amplitude value in the current harmonic characteristic set is greater than a preset amplitude threshold, the analysis frame signal corresponding to the amplitude value greater than the preset amplitude threshold is marked as a harmonic anomaly.

[0085] Select one frame of the power quality data as the target frame one by one;

[0086] Determine whether the target frame contains the voltage anomaly tag or the harmonic anomaly;

[0087] If the target frame does not contain the voltage anomaly label and the harmonic anomaly label, then the target frame is marked as a normal frame;

[0088] Differential coding is performed on the power quality data of all the normal frames to obtain coded quality data;

[0089] The encoded quality data is rapidly compressed to obtain a compressed normal waveform;

[0090] If the target frame contains the voltage anomaly tag or the harmonic anomaly, then the target frame is marked as an anomaly frame;

[0091] Lossless compression is performed on the power quality data of all the aforementioned abnormal frames to obtain compressed abnormal waveforms.

[0092] The compressed normal waveform and the compressed abnormal waveform are combined into a compressed waveform.

[0093] In detail, by determining whether there are voltage anomaly tags or harmonic anomaly markers in the target frame, frames in which voltage and current anomalies occur simultaneously or individually can be quickly identified, thus enabling precise location of power quality anomaly events.

[0094] If the target frame does not contain voltage anomaly tags or harmonic anomalies, then the target frame is a normal frame. For frames marked as normal in the power quality data, the power quality data of each frame is differentially calculated with the power quality data of the previous frame. The difference is used as the encoding result to obtain the encoded quality data, which effectively reduces data redundancy and storage requirements while maintaining the variation characteristics of the power quality data.

[0095] The encoded quality data is divided into blocks of fixed length. Redundancy detection is performed on each block to identify low-amplitude differences or repetition patterns between consecutive frames. Efficient coding methods, such as zero-value compression, entropy coding, or arithmetic coding, are used to compress the low-amplitude differences or repetition patterns. The high-frequency data of the encoded quality data block is mapped to short codewords while retaining key power quality characteristics. The compressed blocks are then combined to generate a compressed normal waveform.

[0096] If the target frame contains a voltage anomaly tag or harmonic anomaly, then the target frame is an abnormal frame.

[0097] Specifically, the lossless compression of power quality data for all the abnormal frames to obtain compressed abnormal waveforms includes:

[0098] Wavelet transform is performed on the power quality data of all the aforementioned abnormal frames to obtain a multi-scale set of wavelet coefficients;

[0099] Calculate the energy distribution of the wavelet coefficient set, and obtain the energy value of each wavelet coefficient based on the energy distribution;

[0100] The energy values ​​are sorted in descending order, and the target wavelet coefficient set corresponding to the energy values ​​that are greater than the preset cumulative energy ratio is selected from the sorting results.

[0101] The target wavelet coefficient set is quantized to obtain a quantized coefficient set;

[0102] Entropy encoding is performed on the quantization coefficient set to obtain the encoded coefficient set;

[0103] Perform an inverse wavelet transform on the set of coding coefficients to obtain a sparse approximate waveform;

[0104] Obtain the original waveforms of the power quality data of all the abnormal frames, and subtract the original waveforms from the sparse approximation waveforms to obtain the residual signal;

[0105] The residual signal is losslessly compressed to obtain the compressed residual;

[0106] The set of coding coefficients and the compression residual are combined to form the compression anomaly waveform.

[0107] In detail, wavelet transform is performed on the voltage and current waveforms of each abnormal frame to decompose the original time-domain signal into multiple frequency band components of different scales, thereby obtaining a multi-scale wavelet coefficient set that reflects the local time-domain and frequency characteristics of the signal.

[0108] For each abnormal frame, the set of wavelet coefficients is used to calculate the energy distribution of coefficients at each scale and frequency band. That is, the energy value of each wavelet coefficient is obtained by squaring and summing the values. This allows for the quantification of the energy distribution characteristics of the signal at different scales. By obtaining the energy value of each wavelet coefficient, the local strength of the abnormal waveform in the time and frequency domain can be reflected.

[0109] The continuous amplitudes in the target wavelet coefficient set are mapped to finite discrete values ​​according to a preset quantization rule, thereby obtaining a quantized coefficient set. Quantization not only reduces the redundancy in the precision of data representation, but also preserves the main characteristics of abnormal waveforms.

[0110] Entropy coding methods (such as Huffman coding or arithmetic coding) are used to efficiently encode the quantization coefficient set, mapping high-frequency coefficients to short codewords and low-frequency coefficients to long codewords, thereby generating a coded coefficient set. This significantly reduces data redundancy while preserving signal characteristics as much as possible, achieving efficient compression of abnormal frame power quality data.

[0111] Based on the scale and mother wavelet function used in wavelet decomposition, the set of coding coefficients is reconstructed in the time domain to generate a sparse approximate waveform, which can remove redundant information while preserving the main features of the power quality signal of abnormal frames.

[0112] The original voltage and current waveforms of all frames marked as anomalous are obtained from power quality data. Each frame's original waveform is then subtracted point-by-point from its corresponding sparse approximation waveform to obtain a residual signal, reflecting high-frequency details and local anomalous features that the sparse approximation failed to preserve. The residual signal is then compressed using a lossless compression algorithm (such as LZ77, DEFLATE, or entropy coding) to minimize data redundancy and storage space while ensuring information integrity. The previously obtained set of coding coefficients (the core features of the sparse approximation waveform) is combined with the compressed residual signal to form a complete compressed anomalous waveform.

[0113] By classifying and compressing power quality data into normal and abnormal frames, targeted data processing is achieved: normal frames utilize differential coding and fast compression to reduce redundancy, enabling efficient storage and transmission; abnormal frames extract multi-scale features through wavelet decomposition, quantization, and entropy coding to generate sparse approximate waveforms, and combine this with lossless compression of residual signals. This preserves the main features and subtle details of the abnormal waveforms while significantly reducing the data volume, ultimately combining the compressed normal and abnormal waveforms into a complete compressed waveform. This classification and compression strategy can significantly improve storage and transmission efficiency while ensuring the integrity of key power quality features, providing an efficient and reliable data foundation for power quality analysis, anomaly detection, and subsequent management.

[0114] S4. Transmit compressed data and feature information through a redundant channel that includes primary and backup communication links. Specifically, a redundant communication channel is established between the IoT power quality sensor and the edge computing gateway of the preset substation. The redundant communication channel includes a 5G wireless link and a fiber optic backup channel.

[0115] In this embodiment of the invention, the edge computing gateway is a local processing node set up in a preset substation. It has data storage, fast analysis and uplink communication functions and is used to preprocess and forward the data collected by the IoT power quality sensor.

[0116] Redundant communication channels: These refer to the simultaneous configuration of two or more independent communication links, supporting primary / backup switching to ensure communication continuity and reliability. 5G wireless link: A high-speed, low-latency wireless data channel based on cellular mobile networks, serving as the default primary link. Fiber optic backup channel: A high-bandwidth channel based on wired fiber optic communication technology, undertaking data transmission tasks when the primary link fails.

[0117] In detail, in AC / DC distribution networks, to ensure stable transmission of power quality data from IoT power quality sensors to the edge computing gateway of a pre-set substation, the system employs a redundant communication channel design: By default, the sensors upload real-time collected power quality data to the edge gateway via a 5G wireless link, achieving high-speed transmission with low latency; simultaneously, the fiber optic backup channel is in standby mode and continuously monitors the link. When interference, excessive latency, or interruption occurs on the 5G wireless link, the system automatically switches to the fiber optic backup channel to ensure uninterrupted and reliable transmission of power quality data and alarm information to the edge computing gateway.

[0118] By establishing redundant communication channels, including a 5G wireless link and a fiber optic backup channel, between the IoT power quality sensor and the substation edge computing gateway, multi-path communication assurance is achieved. When the main link fails or is interfered with, the backup channel can immediately take over the data transmission, ensuring the continuity and reliability of power quality monitoring data. This improves the system's anti-interference capability, reduces the risk of data loss, and enhances the stability and security of power quality monitoring and real-time management.

[0119] S5. Generate an alarm data packet based on the current harmonic characteristic set, the voltage anomaly label, and the compressed waveform, and upload the alarm data packet to the preset dispatch center through the redundant communication channel.

[0120] In this embodiment of the invention, during the operation of the AC / DC distribution network, the alarm data packet not only carries the type and characteristic parameters of the power quality anomaly, but also includes a compressed original waveform as auxiliary verification information. The alarm data packet is transmitted through redundant communication channels: under normal circumstances, the 5G wireless link is used first to achieve low-latency uploading; when the 5G link fails or the latency is too high, the system automatically switches to the fiber optic backup channel to ensure that the alarm data can be reliably and in real time transmitted to the preset dispatch center, thereby supporting the dispatch center to make rapid power quality control decisions.

[0121] Pre-set dispatch center: A centralized management and control platform located in the park or regional power grid, responsible for receiving alarm data, conducting power quality assessment and decision control, and issuing dispatch instructions to energy storage devices, flexible interconnection devices, etc.

[0122] Specifically, the step of generating an alarm data packet based on the current harmonic characteristic set, the voltage anomaly tag, and the compressed waveform, and uploading the alarm data packet to a preset dispatch center through the redundant communication channel, includes:

[0123] The current harmonic feature set, the voltage anomaly label, and the compressed waveform are dynamically weighted to obtain a fused dataset.

[0124] The fused dataset is formatted and encapsulated to obtain an encapsulated dataset;

[0125] A compressed fingerprint is generated based on the compressed waveform, and the compressed fingerprint is embedded into the encapsulated dataset to obtain an updated dataset.

[0126] Obtain the unique identifier of the updated dataset, and generate an alarm data packet based on the unique identifier and the updated dataset;

[0127] The real-time availability of the 5G wireless link in the redundant communication channel is detected, and the detection results are obtained.

[0128] When the detection result indicates that the 5G wireless link is available, the 5G wireless link of the redundant communication channel is selected;

[0129] When the detection result indicates that the 5G wireless link is unavailable, the fiber optic backup channel of the redundant communication channel is selected;

[0130] The alarm data packet is uploaded to the preset dispatch center through the redundant communication channel.

[0131] In detail, the weights are adjusted in real time based on the data importance, anomaly degree or time sensitivity of the current harmonic feature set, voltage anomaly label and compressed waveform. The weighted data is then fused to generate a fused dataset that comprehensively reflects the power quality status.

[0132] The fused dataset is processed according to a unified data structure and protocol. Necessary metadata such as identification fields, timestamps, frame numbers, and check codes are added to various types of information in the fused dataset. Current harmonic characteristics, voltage anomaly labels, and compressed waveforms are then systematically integrated into standardized data units to generate a packaged dataset.

[0133] Feature extraction is performed on the compressed waveform, selecting indicators that represent key waveform characteristics, such as main frequency components, amplitude features, or compression coefficient patterns. Based on these features, a unique compression fingerprint is generated to identify the core information and integrity of the compressed waveform. The generated compression fingerprint is embedded into a specified field of the encapsulated dataset, and metadata (such as timestamps, frame numbers, or verification information) is updated to form an updated dataset.

[0134] Extract unique identifiers from the updated dataset. These unique identifiers can be used to uniquely identify the updated dataset and its corresponding power quality status. Based on the unique identifiers and the content of the updated dataset, key information (such as current harmonic characteristics, voltage anomaly labels, compressed waveforms, and compressed fingerprint codes) is integrated and an alarm data packet is generated according to the alarm protocol.

[0135] Real-time monitoring of 5G wireless links in redundant communication channels is performed. By continuously collecting performance indicators such as link signal strength, latency, and packet loss rate, the current availability of the 5G wireless link is determined and a detection result is generated. When the detection result indicates that the 5G wireless link is available, it is used as the primary transmission channel to send alarm data packets, ensuring low-latency and high-bandwidth real-time transmission. If the detection result indicates that the 5G link is unavailable or its performance degrades to an unacceptable threshold, it automatically switches to the fiber optic backup channel. The stability and high reliability of fiber optics ensure uninterrupted data transmission, and the alarm data packets are uploaded to the preset dispatch center through the selected channel. At the same time, the transmission path and status information are recorded to ensure that power quality alarm information can be continuously and reliably transmitted, achieving high availability and fault tolerance of the monitoring system.

[0136] By dynamically weighting and formatting the current harmonic characteristics, voltage anomaly labels, and compressed waveforms, a comprehensive representation and structured management of power quality data is achieved. By generating compressed fingerprint codes and embedding them into the encapsulated dataset, and simultaneously generating alarm data packets using unique identifiers, rapid and unique identification and integrity verification of abnormal events are realized. Combined with an intelligent selection mechanism for redundant communication channels (5G main link and fiber optic backup channel), alarm data can still be reliably transmitted to the dispatch center even in the event of main link failure or interference. This not only improves the accuracy and response speed of power quality anomaly alarms but also enhances the reliability, fault tolerance, and security of data transmission, providing efficient and robust technical support for real-time power grid monitoring and intelligent control.

[0137] S6. The preset dispatch center that receives the alarm data packet performs decision control on the power quality data to obtain the power quality target management result.

[0138] In this embodiment of the invention, after receiving the alarm data packet from the sensor, the preset dispatch center analyzes the current harmonic characteristics, voltage anomaly tags and compressed waveform information in the packet, and combines historical power quality data and real-time operating parameters to comprehensively evaluate the current state of the power grid. Based on the preset control strategy and reinforcement learning optimization model, it performs decision analysis on possible harmonics, fluctuations and abnormal events, generates targeted adjustment schemes, and thus obtains the power quality target control results.

[0139] In detail, the step of using a preset dispatch center that receives the alarm data packet to perform decision control on the power quality data to obtain the power quality target management result includes:

[0140] Generate an event risk control value for the power quality data based on the alarm data packet;

[0141] The priority of abnormal events in the power quality data is determined based on the event risk control value.

[0142] Based on the event priority, a corresponding candidate control scheme is matched in a preset two-layer knowledge base, and the candidate control scheme is used as the action space.

[0143] Collect the operating status of AC / DC distribution network parameters and use the operating status of these parameters as the state space;

[0144] One candidate control scheme in the action space is selected as the target action, and one parameter running state in the state space is selected as the target state. The target action and the target state are then used as a target analysis group.

[0145] Obtain the state transition probability of the target state in the target analysis group, and the instantaneous reward of the target analysis group;

[0146] A reinforcement learning optimization model is constructed based on the action space, the state space, the state transition probability, the immediate reward, and the preset discount factor.

[0147] The candidate control scheme is iterated using the reinforcement learning optimization model, the number of iterations is counted, and the final cumulative reward is calculated based on the immediate reward and the number of iterations.

[0148] When the final accumulated reward reaches the preset reward threshold, the iteration stops, and the final candidate control scheme is taken as the optimal control scheme.

[0149] The optimal control scheme is converted into control commands, and the control commands are executed on the power quality data to obtain the power quality target management results after execution.

[0150] In detail, after receiving the alarm data packet, the dispatch center analyzes the key information in the alarm data packet, such as the current harmonic characteristic set, voltage anomaly label and compressed waveform. Combined with the timestamp of the abnormal event, the abnormal amplitude and historical statistical patterns, the center quantifies the risks that the abnormal event may cause, such as power grid fluctuations, equipment losses or power outages. The center then calculates the corresponding event risk control value through a weighted risk score or probabilistic statistical model.

[0151] The event risk control value is compared with the preset risk threshold range, and abnormal events in the power quality data are prioritized: if the event risk control value is less than the lower limit of the risk threshold range, the event corresponding to the event risk control value is marked as low priority and only recorded or delayed; if the event risk control value is within the risk threshold range, the event corresponding to the event risk control value is marked as medium priority and enters the normal scheduling and control process; if the event risk control value is greater than the upper limit of the risk threshold range, the event corresponding to the event risk control value is marked as high priority, triggers an emergency alarm, and is given priority in the allocation of governance and scheduling resources.

[0152] The system uses the priority of abnormal events as the search criteria and enters a preset two-layer knowledge base for matching. The first layer of the knowledge base stores general control strategies based on rules and expert experience, while the second layer contains personalized control schemes generated by combining historical data, mechanism models, and intelligent optimization algorithms. The system first quickly locates initial candidate schemes that meet the requirements of event type and priority in the first layer of the knowledge base, and then further filters out more targeted and optimal candidate control schemes in the second layer of the knowledge base.

[0153] All candidate control schemes are denoted as the action space. The operating state of AC / DC distribution network parameters is recorded as the state space. The candidate control schemes and parameter operating states are combined one by one, and the combination is then... As the target analysis group.

[0154] Using the state transition probabilities of the target states in the target analysis group Instant rewards Discount Factor Constructing Markov Decision Processes A reinforcement learning optimization model is constructed using Markov decision processes and policy iteration.

[0155] The policy iteration is performed on the reinforcement learning optimization model to continuously update the decision policy of the candidate control schemes. The calculation formula is shown below:

[0156]

[0157] in, Indicates the parameter running status The following candidate control schemes will be adopted. Number of samplings Indicates the first The immediate reward obtained from the next sample Indicates the parameter running status The following candidate control schemes will be adopted. Instant rewards.

[0158]

[0159] in, Indicates the parameter running status The following candidate control schemes will be adopted. Number of samplings Indicates the parameter running status The following candidate control schemes will be adopted. And switch to parameter running state. Number of times, Indicates the parameter running status The following candidate control schemes will be adopted. And switch to parameter running state. The state transition probability.

[0160]

[0161] in, Indicates the parameter running status The following candidate control schemes will be adopted. strategy, Indicates an immediate reward. Indicates the discount factor. Indicates the parameter running status The following candidate control schemes will be adopted. And switch to parameter running state. The state transition probability, Indicates the parameter running status Next, according to the strategy The expected cumulative reward of the action, Indicates the parameter running status Next, according to the strategy The expected cumulative reward for the action.

[0162]

[0163] in, Indicates the parameter running status The following candidate control schemes will be adopted. Instant rewards Indicates the discount factor. Indicates the parameter running status The following candidate control schemes will be adopted. And switch to parameter running state. The state transition probability, Indicates the parameter running status Next, according to the strategy The expected cumulative reward of the action, Indicates the parameter running status The following candidate control schemes will be adopted. The expected cumulative reward.

[0164]

[0165] in, Indicates the parameter running status The optimal control scheme selected below. Indicates the parameter running status The following candidate control schemes will be adopted. Expected cumulative reward Indicates selection Largest candidate control scheme .

[0166] The number of iterations is counted. During the iteration process, the final cumulative reward is calculated based on the immediate reward obtained in each round and the corresponding number of iterations. When the final cumulative reward reaches the reward threshold, the strategy is considered to have converged, the iteration stops, and the candidate control scheme obtained at this time is determined as the optimal control scheme. The calculation formula is as follows:

[0167]

[0168] in, Indicates the number of iterations. Indicates the first Instant rewards for time steps Indicates the first Discount factor for time step This indicates the number of iterations from the 1st iteration to the 2nd iteration. The final cumulative reward of the next iteration.

[0169] The obtained optimal control scheme is transformed into control commands that can be directly issued and applied to the power quality data of the distribution network. The power quality data is adjusted and optimized in real time, thereby obtaining the power quality target control results after execution, and realizing the effective improvement and optimization control of power quality.

[0170] By hierarchically managing event risk control values ​​and abnormal event priorities, the scientific and rational nature of alarm responses can be ensured. A two-layer knowledge base is used to match candidate control schemes, and a reinforcement learning optimization model is combined to iterate strategies in state-action interactions. This allows control strategies to continuously optimize and gradually converge to the optimal in a dynamic environment, thereby improving the accuracy and adaptability of the control scheme. The optimal control scheme is then transformed into control commands and directly applied to power quality data, achieving closed-loop management from risk perception and strategy optimization to execution feedback. This not only improves the stability and reliability of power quality but also enhances the intelligent management capabilities of the distribution network.

[0171] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0172] like Figure 2 The diagram shown is a functional block diagram of an AC / DC power distribution network power quality control system according to an embodiment of the present invention.

[0173] This disclosure provides an AC / DC distribution network power quality control system, which corresponds one-to-one with the AC / DC distribution network power quality control method described in the above embodiments. For example... Figure 2 As shown, the AC / DC power distribution network power quality control system 100 can be installed in electronic devices. According to its functions, the AC / DC power distribution network power quality control system 100 includes a current harmonic analysis module 101, a voltage anomaly determination module 102, a data classification and compression module 103, a redundant communication channel establishment module 104, an alarm data packet generation module 105, and an alarm data management module 106. Detailed descriptions of each functional module are as follows:

[0174] The current harmonic analysis module 101 is used to collect power quality data from several Internet of Things power quality sensors in the AC / DC power distribution network, perform harmonic analysis on the current waveform of the power quality data, and obtain a set of current harmonic features.

[0175] The voltage anomaly determination module 102 is used to adaptively determine the voltage fluctuation of the power quality data, and filter out the voltage anomaly waveforms according to the determination result to generate voltage anomaly tags.

[0176] The data classification and compression module 103 is used to classify and compress the power quality data according to the current harmonic feature set and the voltage anomaly label to obtain a compressed waveform.

[0177] The redundant communication channel establishment module 104 is used to establish a redundant communication channel between the IoT power quality sensor and the edge computing gateway of the preset substation. The redundant communication channel includes a 5G wireless link and a fiber optic backup channel.

[0178] The alarm data packet generation module 105 is used to generate an alarm data packet based on the current harmonic feature set, the voltage anomaly label and the compressed waveform, and upload the alarm data packet to the preset dispatch center through the redundant communication channel;

[0179] The alarm data management module 106 is used to make decisions and controls on the power quality data by using a preset dispatch center that receives the alarm data packet, so as to obtain the power quality target management result.

[0180] In one embodiment, the current harmonic analysis module 101 performs harmonic analysis on the current waveform of the power quality data to obtain a set of current harmonic features, including:

[0181] The power quality data is timestamped to obtain aligned quality data;

[0182] The alignment quality data is subjected to noise removal using a preset adaptive filter to obtain denoised quality data.

[0183] The denoised quality data is normalized to obtain normalized quality data;

[0184] Extract the current waveform from the normalized quality data;

[0185] Obtain the waveform division length, window function, and overlap rate, and divide the current waveform into several initial frame signals according to the waveform division length;

[0186] The initial frame signal is optimized using the window function and the overlap rate to obtain the analysis frame signal;

[0187] Perform a Fast Fourier Transform on the analyzed frame signal to obtain the amplitude spectrum and phase spectrum for each frequency;

[0188] The fundamental frequency is determined from the amplitude spectrum and the phase spectrum, and the fundamental amplitude and fundamental phase of the fundamental frequency are identified.

[0189] Obtain the integer harmonics of the fundamental frequency, and extract the harmonic amplitude and phase of the integer harmonics.

[0190] The harmonic content of each integer octave is calculated based on the fundamental frequency amplitude and the octave amplitude.

[0191] Calculate the total harmonic distortion rate based on the harmonic content rate;

[0192] The fundamental amplitude, the fundamental phase, the octave amplitude, the octave phase, the harmonic content, and the total harmonic distortion rate are used to form a current harmonic characteristic set.

[0193] In one embodiment, the voltage anomaly determination module 102 performs adaptive threshold determination on the voltage fluctuations of the power quality data, and filters out voltage anomaly waveforms based on the determination result to generate voltage anomaly tags, including:

[0194] Obtain the historical voltage waveform of historical power quality data, and calculate the historical voltage mean and historical voltage standard deviation based on the historical voltage waveform;

[0195] Extract the voltage waveform of the normalized quality data and obtain the instantaneous voltage value of the voltage waveform in each frame;

[0196] The fluctuation amplitude of the voltage waveform in each frame is calculated based on the historical average voltage value and the instantaneous voltage value;

[0197] A dynamic fluctuation threshold range is generated using the historical voltage mean, the historical voltage standard deviation, and a preset fluctuation factor.

[0198] Determine whether the fluctuation amplitude is greater than or equal to the lower limit of the dynamic fluctuation threshold range, and less than or equal to the upper limit of the dynamic fluctuation threshold range;

[0199] If the fluctuation amplitude is greater than or equal to the lower limit of the dynamic fluctuation threshold range and less than or equal to the upper limit of the dynamic fluctuation threshold range, then the voltage waveform corresponding to the fluctuation amplitude is determined to be a normal voltage waveform.

[0200] If the fluctuation amplitude is less than the lower limit of the dynamic fluctuation threshold range or greater than the upper limit of the dynamic fluctuation threshold range, the voltage waveform corresponding to the fluctuation amplitude is determined to be a voltage abnormal waveform, and a voltage abnormality label is generated based on the fluctuation amplitude.

[0201] In one embodiment, the data classification and compression module 103 performs classification and compression of the power quality data based on the current harmonic feature set and the voltage anomaly label to obtain a compressed waveform, including:

[0202] When any amplitude value of the current harmonic feature set is greater than a preset amplitude threshold, the analysis frame signal corresponding to the amplitude value greater than the preset amplitude threshold is marked as a harmonic anomaly.

[0203] Select one frame of the power quality data as the target frame one by one;

[0204] Determine whether the target frame contains the voltage anomaly tag or the harmonic anomaly;

[0205] If the target frame does not contain the voltage anomaly label and the harmonic anomaly label, then the target frame is marked as a normal frame;

[0206] Differential coding is performed on the power quality data of all the normal frames to obtain coded quality data;

[0207] The encoded quality data is rapidly compressed to obtain a compressed normal waveform;

[0208] If the target frame contains the voltage anomaly tag or the harmonic anomaly, then the target frame is marked as an anomaly frame;

[0209] Lossless compression is performed on the power quality data of all the aforementioned abnormal frames to obtain compressed abnormal waveforms.

[0210] The compressed normal waveform and the compressed abnormal waveform are combined into a compressed waveform.

[0211] In one embodiment, the data classification and compression module 103 performs classification and compression of the power quality data based on the current harmonic feature set and the voltage anomaly label to obtain a compressed waveform, including:

[0212] Wavelet transform is performed on the power quality data of all the aforementioned abnormal frames to obtain a multi-scale set of wavelet coefficients;

[0213] Calculate the energy distribution of the wavelet coefficient set, and obtain the energy value of each wavelet coefficient based on the energy distribution;

[0214] The energy values ​​are sorted in descending order, and the target wavelet coefficient set corresponding to the energy values ​​that are greater than the preset cumulative energy ratio is selected from the sorting results.

[0215] The target wavelet coefficient set is quantized to obtain a quantized coefficient set;

[0216] Entropy encoding is performed on the quantization coefficient set to obtain the encoded coefficient set;

[0217] Perform an inverse wavelet transform on the set of coding coefficients to obtain a sparse approximate waveform;

[0218] Obtain the original waveforms of the power quality data of all the abnormal frames, and subtract the original waveforms from the sparse approximation waveforms to obtain the residual signal;

[0219] The residual signal is losslessly compressed to obtain the compressed residual;

[0220] The set of coding coefficients and the compression residual are combined to form the compression anomaly waveform.

[0221] In one embodiment, the alarm data packet generation module 105 generates an alarm data packet based on the current harmonic feature set, the voltage anomaly tag, and the compressed waveform, and uploads the alarm data packet to a preset dispatch center through the redundant communication channel, including:

[0222] The current harmonic feature set, the voltage anomaly label, and the compressed waveform are dynamically weighted to obtain a fused dataset.

[0223] The fused dataset is formatted and encapsulated to obtain an encapsulated dataset;

[0224] A compressed fingerprint is generated based on the compressed waveform, and the compressed fingerprint is embedded into the encapsulated dataset to obtain an updated dataset.

[0225] Obtain the unique identifier of the updated dataset, and generate an alarm data packet based on the unique identifier and the updated dataset;

[0226] The real-time availability of the 5G wireless link in the redundant communication channel is detected, and the detection result is obtained;

[0227] When the detection result indicates that the 5G wireless link is available, the 5G wireless link of the redundant communication channel is selected;

[0228] When the detection result indicates that the 5G wireless link is unavailable, the fiber optic backup channel of the redundant communication channel is selected;

[0229] The alarm data packet is uploaded to the preset dispatch center through the redundant communication channel.

[0230] In one embodiment, the alarm data management module 106 performs decision control on the power quality data using a preset dispatch center that receives the alarm data packet, and obtains the power quality target management result, including:

[0231] Generate an event risk control value for the power quality data based on the alarm data packet;

[0232] The priority of abnormal events in the power quality data is determined based on the event risk control value.

[0233] Based on the event priority, a corresponding candidate control scheme is matched in a preset two-layer knowledge base, and the candidate control scheme is used as the action space.

[0234] Collect the operating status of AC / DC distribution network parameters and use the operating status of these parameters as the state space;

[0235] One candidate control scheme in the action space is selected as the target action, and one parameter running state in the state space is selected as the target state. The target action and the target state are then used as a target analysis group.

[0236] Obtain the state transition probability of the target state in the target analysis group, and the instantaneous reward of the target analysis group;

[0237] A reinforcement learning optimization model is constructed based on the action space, the state space, the state transition probability, the immediate reward, and the preset discount factor.

[0238] The candidate control scheme is iterated using the reinforcement learning optimization model, the number of iterations is counted, and the final cumulative reward is calculated based on the immediate reward and the number of iterations.

[0239] When the final accumulated reward reaches the preset reward threshold, the iteration stops, and the final candidate control scheme is taken as the optimal control scheme.

[0240] The optimal control scheme is converted into control commands, and the control commands are executed on the power quality data to obtain the power quality target management results after execution.

[0241] This invention adaptively thresholds voltage fluctuations in power quality data and filters out abnormal voltage waveforms based on the thresholds, generating voltage anomaly tags. It does not rely on fixed thresholds, allowing the thresholds to dynamically adjust with the power grid's operating status, thereby improving the accuracy and sensitivity of voltage anomaly detection. Power quality data is classified and compressed based on the current harmonic feature set and the voltage anomaly tags to obtain compressed waveforms. Normal frames utilize differential coding and fast compression to reduce redundancy, achieving efficient storage and transmission. Abnormal frames extract multi-scale features through wavelet decomposition, quantization, and entropy coding to generate sparse approximate waveforms. Combined with lossless compression of residual signals, this preserves the main features and subtle details of the abnormal waveforms while significantly reducing data volume. The compressed normal waveforms and compressed abnormal waveforms are combined into a complete compressed waveform. A redundant communication channel is established between the IoT power quality sensor and the edge computing gateway of a preset substation. This redundant communication channel includes a 5G wireless link and a fiber optic backup channel. Based on the current harmonics... The feature set, the voltage anomaly label, and the compressed waveform generate an alarm data packet, which is then uploaded to a preset dispatch center via the redundant communication channel. Combined with the intelligent selection mechanism of the redundant communication channel (5G main link and fiber optic backup channel), this ensures that alarm data can still be reliably transmitted to the dispatch center even in the event of main link failure or interference. This not only improves the accuracy and response speed of power quality anomaly alarms but also enhances the reliability, fault tolerance, and security of data transmission. The preset dispatch center, upon receiving the alarm data packet, performs decision-making control on the power quality data. By combining a reinforcement learning optimization model with state-action interaction, the control strategy can be continuously optimized and gradually converge to the optimal value in a dynamic environment. This improves the accuracy and adaptability of the control scheme, yields the power quality target management results, and enhances monitoring coverage, real-time performance, and reliability. This effectively meets the high-precision, high-reliability, and fast-response power quality management requirements of modern AC / DC hybrid distribution networks.

[0242] For specific limitations regarding an AC / DC distribution network power quality control system, please refer to the limitations of an AC / DC distribution network power quality control method described above, which will not be repeated here. Each module in the aforementioned AC / DC distribution network power quality control system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0243] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps:

[0244] Power quality data from several IoT power quality sensors in an AC / DC power distribution network are collected, and harmonic analysis is performed on the current waveform of the power quality data to obtain a set of current harmonic characteristics.

[0245] An adaptive threshold is used to determine the voltage fluctuation of the power quality data, and voltage anomaly waveforms are selected based on the determination results to generate voltage anomaly tags.

[0246] The power quality data is classified and compressed based on the current harmonic feature set and the voltage anomaly label to obtain a compressed waveform.

[0247] A redundant communication channel is established between the IoT power quality sensor and the edge computing gateway of the preset substation. The redundant communication channel includes a 5G wireless link and a fiber optic backup channel.

[0248] An alarm data packet is generated based on the current harmonic characteristic set, the voltage anomaly label, and the compressed waveform, and the alarm data packet is uploaded to the preset dispatch center through the redundant communication channel;

[0249] The preset dispatch center that receives the alarm data packet performs decision control on the power quality data to obtain the power quality target management result.

[0250] In the embodiments provided by this invention, it should be understood that the disclosed devices and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0251] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0252] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within the invention. No appended diagram markings in the claims should be construed as limiting the scope of the claims.

[0253] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0254] In some embodiments of this example, a computer-readable storage medium is provided, on which a computer program is stored, characterized in that the computer program, when executed by a processor, implements the steps of the method described in the above embodiments.

[0255] The readable storage medium of the present invention stores a computer program, which, when executed by a processor of an electronic device, can perform the following:

[0256] Power quality data from several IoT power quality sensors in an AC / DC power distribution network are collected, and harmonic analysis is performed on the current waveform of the power quality data to obtain a set of current harmonic characteristics.

[0257] An adaptive threshold is used to determine the voltage fluctuation of the power quality data, and voltage anomaly waveforms are selected based on the determination results to generate voltage anomaly tags.

[0258] The power quality data is classified and compressed based on the current harmonic feature set and the voltage anomaly label to obtain a compressed waveform.

[0259] A redundant communication channel is established between the IoT power quality sensor and the edge computing gateway of the preset substation. The redundant communication channel includes a 5G wireless link and a fiber optic backup channel.

[0260] An alarm data packet is generated based on the current harmonic characteristic set, the voltage anomaly label, and the compressed waveform, and the alarm data packet is uploaded to the preset dispatch center through the redundant communication channel;

[0261] The preset dispatch center that receives the alarm data packet performs decision control on the power quality data to obtain the power quality target management result.

[0262] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or computer device described above can be referred to the relevant descriptions on the server side and client side in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0263] Computer-readable storage media may also store at least one computer-executable program / instruction, such as computer-readable instructions. Computer-readable storage media include, but are not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory. Computer-readable storage media may include, for example, read-only memory (ROM), hard disk, flash memory, etc. For example, a non-transitory computer-readable storage medium may be connected to a computing device such as a computer, and then, when the computing device executes the computer-readable instructions stored on the computer-readable storage medium, the various methods described above can be performed.

[0264] In addition, the computer device may include (but is not limited to) a data bus, an input / output (I / O) bus, a display, and input / output devices (e.g., keyboard, mouse, speakers, etc.).

[0265] The processor can communicate with external devices via the I / O bus through wired or wireless networks.

[0266] In one embodiment, the at least one computer-executable instruction may also be compiled into or comprise a software product / computer program product, wherein one or more computer-executable instructions are executed by a processor to perform the steps of the various functions and / or methods in the embodiments described herein.

[0267] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0268] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above.

[0269] In the embodiments provided in this disclosure, it should be understood that the disclosed systems and methods can also be implemented in other ways. The system embodiments described above are merely illustrative; for example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0270] It should be noted that, in this disclosure, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element limited by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0271] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A method for power quality control in AC / DC distribution networks, characterized in that, The method includes: Collect power quality data from AC / DC distribution networks, perform feature analysis on the power quality data, and obtain current characteristic parameters and voltage anomaly indicators; Based on current characteristic parameters and voltage anomaly indicators, power quality data is subjected to differential compression processing; Compressed data and feature information are transmitted through redundant channels that include primary and backup communication links; Power quality alarm information is generated based on the transmitted data, and power grid operation decision-making and control are carried out based on this information; Differential compression processing is performed on power quality data, including: Frame by frame, determine whether the target frame in the power quality data is marked as voltage abnormality or harmonic abnormality. Harmonic abnormality is obtained by comparing the amplitude of the current harmonic characteristics with the preset amplitude threshold. If the target frame has no voltage or harmonic abnormalities, it is marked as a normal frame, and differential coding and lossy compression are performed on it to obtain a compressed normal waveform. If any anomaly exists in the target frame, it is marked as an anomalous frame and subjected to lossless compression to obtain a compressed anomalous waveform; all compressed normal waveforms and compressed anomalous waveforms are summarized to generate a compressed waveform. The specific method for obtaining compressed abnormal waveforms is as follows: Wavelet transform is performed on abnormal frame power quality data to obtain a multi-scale set of wavelet coefficients; Based on the energy distribution of wavelet coefficient sets, a set of significant coefficients with significant energy is selected; The set of significant coefficients is quantized and losslessly encoded to obtain the set of encoded coefficients. Perform an inverse transform based on the coding coefficients to generate a sparse approximate waveform, and calculate the residual signal between the original waveform and the sparse approximate waveform. The residual signal is losslessly compressed to obtain the compressed residual; The coding coefficients are combined with the compression residuals to form a compression anomaly waveform.

2. The AC / DC distribution network power quality control method as described in claim 1, characterized in that, The feature analysis includes: Harmonic analysis is performed on the current waveform to extract harmonic characteristic parameters; Adaptive threshold analysis is performed on voltage fluctuations to generate voltage anomaly indicators.

3. The AC / DC distribution network power quality control method as described in claim 2, characterized in that, Differentiated compression processing includes: Based on current characteristic parameters and voltage anomaly indicators, anomaly status judgment is performed on each data frame in the power quality data. According to the judgment results, a lossy compression strategy is adopted for normal data frames and a lossless compression strategy is adopted for abnormal data frames. The compressed data are then merged and output.

4. The AC / DC power quality control method for distribution networks as described in claim 2, characterized in that, The current waveform is subjected to harmonic analysis to extract harmonic characteristic parameters, specifically: Power quality data is preprocessed to obtain standardized current waveform signals; The standardized current waveform signal is divided into frames to obtain the analysis frame signal; Perform spectral analysis on the analysis frame signal to obtain the amplitude and phase parameters of the fundamental frequency and its harmonic frequencies; The content of each harmonic and the total harmonic distortion rate are calculated based on the amplitude and phase parameters.

5. The AC / DC power quality control method for distribution networks as described in claim 4, characterized in that, The adaptive threshold analysis of voltage fluctuations to generate voltage anomaly indicators specifically involves: Acquire historical voltage waveforms and calculate the historical voltage mean and standard deviation. Real-time acquisition of voltage waveforms and extraction of instantaneous voltage values; The fluctuation amplitude is calculated based on the historical average voltage and the instantaneous voltage value; Based on the historical voltage mean, historical voltage standard deviation and preset fluctuation factor, a dynamic fluctuation threshold range is generated. Determine whether the fluctuation amplitude is within the range of the dynamic fluctuation threshold; If the voltage is within the range, it is determined to be a normal voltage waveform; if it is outside the range, it is determined to be an abnormal voltage waveform and a voltage abnormality label is generated.

6. The AC / DC distribution network power quality control method as described in claim 1, characterized in that, The step of generating power quality alarm information based on transmitted data and making power grid operation decision-making and control based on this information includes: The current characteristic parameters, voltage anomaly indicators, and compressed waveforms are weighted and fused to generate a fused dataset. The fused dataset is formatted and encapsulated, and a compressed fingerprint code generated based on the compressed waveform is embedded to obtain an updated dataset, which in turn generates an alarm data packet containing the updated dataset and its unique identifier. The system detects the availability of the primary wireless link in the redundant communication channels and adaptively selects either the primary wireless link or the backup wired channel based on the detection results; alarm data packets are then uploaded to the dispatch center through the selected channel.

7. The AC / DC power quality control method for distribution networks as described in claim 1 or 6, characterized in that, The step of generating power quality alarm information based on transmitted data and making power grid operation decision-making and control based on the information further includes: Calculate the event risk control value based on the alarm data packet to determine the priority of abnormal events; Based on priority, candidate control schemes are matched from a preset knowledge base to form an action space; Collect real-time operating parameters of the power distribution network to construct a state space; A reinforcement learning optimization model is constructed based on the action space, state space, and state transition probabilities. The model is used to iterate on candidate control schemes. The iteration stops when the cumulative reward reaches a preset threshold, and the optimal control scheme is output. The optimal control scheme is then converted into control commands and executed to obtain the power quality management results.

8. An electronic device, characterized in that, The electronic device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform a power quality control method for AC / DC distribution networks as described in any one of claims 1 to 7.

Images