Electric energy quality prediction and governance platform based on digital twin technology

Abstract

The invention discloses an electric energy quality prediction and management platform based on a digital twin technology, which relates to the technical field of power system monitoring and comprises a physical space layer, a data space layer, a model space layer and an application space layer. The physical space layer is provided with an electric energy quality monitoring device and an electrical equipment state monitoring device, output multi-time-scale data are transmitted to the data space layer to be synchronously collected and preprocessed, the preprocessed data are input into the model space layer, a physical entity mapping relation is constructed through a digital twin basic model, and the physical entity mapping relation is obtained. The attribute expansion model constructs dynamic feature mapping, the multi-scale analysis model realizes dynamic analysis, and an analysis result is input into the application space layer to execute prediction, early warning and treatment. According to the method, a hierarchical digital twinborn system of'basic mapping-feature extension-dynamic analysis' is established, collaborative analysis of transient disturbance, short-term fluctuation and steady-state features is realized, and effective support is provided for optimal configuration of an electric energy quality management device.

Description

Technical Field

[0001] The present invention relates to the technical field of power system monitoring and control, and in particular to a power quality prediction and management platform based on digital twin technology. Background Art

[0002] As the scale of power systems expands and the types of power loads become increasingly complex, the impact of power quality issues on the safe and stable operation of distribution networks has become increasingly prominent. Power quality issues represented by harmonic pollution, three-phase imbalance, voltage fluctuations and transient disturbances not only affect the normal operation of power supply equipment, but also cause significant economic losses. Digital twin technology provides a new approach to solving power quality monitoring and governance issues by constructing digital mappings of physical entities. It is currently widely used in distribution network operation monitoring, fault diagnosis and predictive maintenance.

[0003] In practical applications, power quality monitoring and control technologies face three main challenges: First, power quality problems include phenomena at multiple time scales, such as transient disturbances, short-term fluctuations, and steady-state characteristics. Existing monitoring systems often use independent analysis methods for different time scales, lack the ability to collaboratively process multi-time scale data, and find it difficult to fully grasp and predict the evolution of power quality problems; second, existing digital twin models mainly focus on the static topological mapping of physical entities, and lack analysis of dynamic characteristics and historical evolution laws, and cannot accurately reflect the dynamic changes in the system's operating status; third, the formulation of power quality control plans lacks a scientific basis, and the type selection, installation location determination, and capacity configuration of compensation devices mainly rely on experience and judgment, making it difficult to achieve precise control.

[0004] Therefore, it is urgent to develop a power quality prediction and management platform based on digital twin technology. By establishing a multi-time scale collaborative analysis mechanism, building a multi-level digital twin mapping system, and designing intelligent management decision-making methods, the accuracy of power quality monitoring, the foresight of prediction, and the effectiveness of management can be improved, providing strong support for ensuring the safe and stable operation of the distribution network. Summary of the invention

[0005] The problem to be solved by the present invention is how to provide a power quality prediction and management platform based on digital twin technology to achieve accurate monitoring, prediction and management of power quality. In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0006] A power quality prediction and management platform based on digital twin technology, characterized by including a physical space layer, a data space layer, a model space layer and an application space layer:

[0007] The physical space layer is used to deploy power quality monitoring devices and electrical equipment status monitoring devices to collect voltage transients, voltage fluctuations, harmonics, three-phase imbalance, and equipment operating status data;

[0008] The data space layer performs data collection and preprocessing at multiple time scales, including the collection and preprocessing of transient disturbance data, short-term fluctuation data, and steady-state characteristic data;

[0009] The model space layer includes the digital twin basic model, the attribute extension model and the multi-scale analysis model. The digital twin basic model is used to construct the physical entity mapping of the distribution system, the attribute extension model is used to construct the dynamic feature mapping based on the digital twin basic model, and the multi-scale analysis model is used to realize the dynamic analysis of the distribution system based on the data collected in the data space layer.

[0010] The application space layer includes a power quality prediction unit, a hierarchical early warning unit and a governance decision unit, wherein the power quality prediction unit is used to predict the changing trend of harmonic distortion rate and three-phase imbalance, the hierarchical early warning unit is used to perform real-time, short-term and trend early warnings, and the governance decision unit is used to formulate emergency compensation and preventive deployment plans.

[0011] A further technical solution is that the physical space layer includes: the power quality monitoring device, including a voltage transient monitoring device, a voltage fluctuation monitoring device, a harmonic monitoring device and a three-phase unbalance monitoring device, wherein the voltage transient monitoring device is used to collect voltage sag, voltage swell and voltage interruption data, the voltage fluctuation monitoring device is used to collect voltage fluctuation and voltage flicker data, the harmonic monitoring device is used to collect harmonic distortion data, and the three-phase unbalance monitoring device is used to collect three-phase voltage unbalance data;

[0012] The electrical equipment status monitoring device is used to collect the transformer load rate, the distribution line load current and the operating status of the switchgear.

[0013] A further technical solution is that the data collection and preprocessing at multiple time scales includes:

[0014] Transient disturbance data collection, using a sampling frequency of more than 10kHz, is used to monitor voltage sags, swells and interruption events. High-frequency sampling is performed through the voltage transient monitoring device, and the collected data is decomposed by wavelet transform to extract transient disturbance characteristics, including voltage sag amplitude, duration and waveform distortion;

[0015] Short-term fluctuation data collection, using minute-level sampling frequency, is used for voltage fluctuation and flicker monitoring. Periodic sampling is performed through the voltage fluctuation monitoring device. After variance analysis of the collected data, short-term fluctuation characteristics are extracted, including fluctuation amplitude and fluctuation frequency.

[0016] Steady-state characteristic data collection, using hourly sampling frequency, is used for monitoring harmonics, three-phase imbalance and load changes. Regular sampling is performed through the harmonic monitoring device, three-phase imbalance monitoring device and electrical equipment status monitoring device. After the collected data is analyzed, steady-state characteristics are extracted, including harmonic spectrum, imbalance and load characteristics;

[0017] The multi-time scale data is spatially synchronized with the clocks of each monitoring device through time synchronization technology to achieve synchronous collection, and is temporally preprocessed according to a sampling strategy that combines an event triggering method with a fixed period method to form a feature data set for different analysis targets.

[0018] A further technical solution is that the physical entity mapping constructed by the digital twin basic model includes:

[0019] A monitored power supply equipment mapping, used to map the physical connection relationship between transformers, distribution lines and switchgear in the physical space layer;

[0020] Grid topology mapping, used to map the bus node connection relationship and the direction of transmission and distribution lines between the monitored power supply equipment;

[0021] Monitoring equipment mapping is used to map the installation locations of the various monitoring devices on the monitored power supply equipment and the collection scope of their monitoring data.

[0022] A further technical solution is that the dynamic feature mapping constructed by the attribute extension model includes:

[0023] Power supply equipment feature mapping, used to construct the rated parameters of the monitored power supply equipment, including transformer rated capacity and short-circuit impedance, distribution line impedance parameters, and rated capacity of switchgear;

[0024] Operation state mapping, used to map the transient disturbance characteristics, short-term fluctuation characteristics and steady-state characteristics collected by the monitoring device onto the digital twin basic model;

[0025] Load characteristic mapping uses the load data and historical operation data collected by the electrical equipment status monitoring device to construct the capacity, power factor and daily load curve of the load carried by the transformer and distribution line.

[0026] A further technical solution is that the multi-scale analysis model specifically includes:

[0027] A transient disturbance analysis model, inputting the transient disturbance characteristics into the model, calculating the disturbance propagation path through the structural relationship of the power grid topology mapping, and calculating the disturbance source location in combination with the impedance parameters of the power supply equipment characteristic mapping;

[0028] A short-term fluctuation analysis model is used to correlate the short-term fluctuation characteristics with the daily load curve characteristics mapped by the load characteristics, and calculate the intensity and attenuation law of the fluctuation source;

[0029] The steady-state characteristic analysis model compares and analyzes the steady-state characteristic with the steady-state characteristic of the operating state mapping, and calculates the distortion source distribution and the unbalanced propagation characteristics.

[0030] A further technical solution is that the application space layer includes a power quality prediction unit, a hierarchical early warning unit and a governance decision unit;

[0031] Among them, the power quality prediction unit predicts the changing trends of harmonic distortion rate, three-phase imbalance and power factor in the distribution system through a shallow neural network model, based on the distortion source distribution and unbalanced propagation characteristics output by the steady-state characteristic analysis model and the daily load curve mapped by the load characteristics.

[0032] A further technical solution is that the hierarchical early warning unit includes four modules:

[0033] The transient warning module receives transient disturbance characteristics and triggers millisecond-level warnings when it detects that the transient disturbance exceeds the preset threshold;

[0034] The short-term warning module receives short-term fluctuation characteristics and triggers a warning when it detects that the voltage fluctuation exceeds a preset threshold and lasts for more than a preset time;

[0035] The steady-state warning module receives steady-state characteristics and triggers a warning when it detects that the harmonic content or imbalance exceeds the standard limit;

[0036] The trend warning module receives the quality degradation trend prediction of the power quality prediction unit and triggers a warning when the predicted value of the power quality indicator exceeds the standard limit.

[0037] A further technical solution is that the governance decision-making unit includes two modes: emergency response and preventive deployment:

[0038] In the emergency response mode, for transient warning, short-term warning and steady-state warning signals, the optimal installation position and required compensation capacity of the dynamic voltage recovery device are determined by using the disturbance source location and propagation path determined by the transient disturbance analysis model in combination with the transient disturbance characteristics. The optimal installation position, required compensation capacity and number of devices of the static VAR generator are determined by using the fluctuation source intensity and attenuation law determined by the short-term fluctuation analysis model in combination with the short-term fluctuation characteristics. The optimal installation position, required compensation capacity and number of devices of the active filter are determined by using the distortion source distribution and unbalanced propagation characteristics determined by the steady-state characteristic analysis model in combination with the steady-state characteristics.

[0039] The preventive deployment mode identifies potential quality problem areas based on the harmonic distortion rate and three-phase imbalance trend predicted by the power quality prediction unit. Combined with the distortion source distribution and imbalance propagation characteristics of the steady-state characteristic analysis model, the installation location and capacity requirements of the compensation device in the problem area are determined, and a long-term configuration planning plan is formulated. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is the core architecture diagram of the power quality prediction and management platform.

[0041] Figure 2 It is the multi-time scale data processing flow of the data space layer.

[0042] Figure 3 This is the model space layer structure diagram.

[0043] Figure 4 This is the functional structure diagram of the application space layer. DETAILED DESCRIPTION

[0044] As embodiment 1 of the present invention

[0045] See also Figure 1 The present invention provides a power quality prediction and management platform based on digital twin technology, including a physical space layer, a data space layer, a model space layer and an application space layer. The layers are connected and interacted through data streams.

[0046] The physical space layer is used as the underlying architecture, and is deployed with power quality monitoring devices and electrical equipment status monitoring devices. Among them, the power quality monitoring devices include voltage transient monitoring devices, voltage fluctuation monitoring devices, harmonic monitoring devices and three-phase unbalance monitoring devices, which are used to collect different types of power quality parameters. The electrical equipment status monitoring device is used to collect the operating status data of transformers, distribution lines and switchgear.

[0047] The data space layer performs data collection and preprocessing based on different time scales. It is divided into three categories according to the sampling frequency: transient disturbance data uses high-frequency sampling above 10kHz, short-term fluctuation data uses minute-level sampling, and steady-state feature data uses hour-level sampling. After preprocessing, the collected data forms feature data sets for different analysis targets.

[0048] The model space layer is the core of this system and includes three main models: the digital twin basic model is used to construct the mapping relationship of the physical entities of the distribution system; the attribute extension model constructs dynamic feature mapping on the basic model; and the multi-scale analysis model realizes system dynamic analysis based on the collected data.

[0049] The application space layer is the top layer of the system, which includes three functional units: the power quality prediction unit is used to predict the changing trend of harmonic distortion rate and three-phase imbalance, the hierarchical early warning unit realizes real-time, short-term and trend early warning, and the governance decision-making unit is responsible for formulating emergency compensation and preventive deployment plans.

[0050] This embodiment realizes a complete chain from data collection, processing, analysis to application through a four-layer architecture design. The functions of each layer are clear and coordinated with each other, providing a systematic solution for the prediction and management of power quality. The system adopts digital twin technology, which improves the accuracy of power quality monitoring and analysis through the deep integration of physical space and digital space.

[0051] As the second embodiment of the present invention

[0052] This embodiment describes the physical space layer in detail based on the first embodiment.

[0053] The physical space layer mainly includes two parts: power quality monitoring device and electrical equipment status monitoring device:

[0054] The power quality monitoring device includes:

[0055] Voltage transient monitoring device: adopts high-speed sampling technology, with a sampling frequency of more than 10kHz, to collect voltage sag, voltage swell, and voltage interruption data. The device adopts a cache mechanism to capture transient events in real time.

[0056] Voltage fluctuation monitoring device: It uses minute-level sampling frequency to collect voltage fluctuation and flicker data. The device obtains the voltage fluctuation amplitude and frequency through continuous sampling calculation.

[0057] Harmonic monitoring device: uses FFT analysis technology to collect harmonic distortion data, and preferably can measure the 2nd to 50th harmonic content.

[0058] Three-phase imbalance monitoring device: used to collect three-phase voltage imbalance data and obtain the imbalance degree through real-time calculation.

[0059] The electrical equipment condition monitoring device mainly monitors three types of objects:

[0060] Transformer monitoring: real-time collection of transformer load rate, oil temperature, and winding temperature operating parameters.

[0061] Distribution line monitoring: Collect line load current, voltage, and power factor operating data.

[0062] Switchgear monitoring: collecting switch opening and closing status, operation times, and mechanical characteristics operating status.

[0063] This embodiment realizes comprehensive monitoring of power quality and equipment operating status by reasonably configuring various monitoring devices, and provides a reliable data basis for multi-time scale analysis of the data space layer.

[0064] As the third embodiment of the present invention

[0065] See also Figure 2 , based on the first and second embodiments, this embodiment describes in detail the multi-time scale data collection and preprocessing of the data space layer.

[0066] The multi-time scale data acquisition and preprocessing of the data space layer includes:

[0067] Transient disturbance data collection: Use a sampling frequency of more than 10kHz and perform high-frequency sampling through a voltage transient monitoring device. The collected data is decomposed by wavelet transform using the db4 wavelet basis function to extract three types of transient disturbance features: the voltage sag amplitude is expressed in per-unit value, the duration is recorded in milliseconds, and the waveform distortion is quantified by the distortion metric. This type of data collection is mainly for transient disturbance events.

[0068] Short-term fluctuation data collection: Use minute-level sampling frequency and perform periodic sampling through the voltage fluctuation monitoring device. The collected data is first subjected to moving window variance analysis. Preferably, the window length is set to 5 minutes and the step length is 1 minute. Two types of short-term fluctuation characteristics are extracted through variance analysis: the fluctuation amplitude records the maximum percentage of voltage deviation from the nominal value, and the fluctuation frequency counts the number of times the voltage fluctuation exceeds the limit value per unit time. This type of data collection is suitable for voltage fluctuation monitoring on a short time scale.

[0069] Steady-state characteristic data collection: hourly sampling frequency is adopted, and regular sampling is performed through harmonic monitoring devices, three-phase imbalance monitoring devices and electrical equipment status monitoring devices. Fourier analysis, symmetrical component decomposition and statistical analysis are performed on the collected data to extract three types of steady-state characteristics: preferably, the harmonic spectrum records the 2-50th harmonic content, the imbalance is calculated by the negative sequence component, and the load characteristics include two indicators: power factor and load rate. This type of data collection is aimed at the steady-state operation characteristics of the system.

[0070] The multi-time scale data is spatially synchronized with each monitoring device through time synchronization technology to achieve synchronous acquisition. Preferably, GPS time synchronization technology is used, and the synchronization accuracy is better than 1 microsecond. A double sampling strategy is adopted in time. Preferably, triggered sampling is used for transient events, and the trigger condition is that the voltage exceeds ±10% of the nominal value; fixed time interval sampling is used for periodic monitoring to form feature data sets for different analysis targets.

[0071] This embodiment uses a multi-time scale data collection and preprocessing solution and adopts different sampling strategies for different types of power quality problems, thereby achieving accurate collection and efficient processing of power quality data.

[0072] As the fourth embodiment of the present invention

[0073] See also Figure 3 , based on the first, second and third embodiments, this embodiment describes the three models of the model space layer in detail.

[0074] The physical entity mapping constructed by the digital twin basic model includes:

[0075] Mapping of monitored power supply equipment: used to map the physical connection relationship between transformers, distribution lines and switchgear in the physical space layer. During the mapping process, the system first defines the transformer as a node element, the distribution line as a connection element, and the switchgear as a control element, and establishes the physical topology structure between the devices.

[0076] Grid topology mapping: used to map the busbar node connection relationship and the direction of the transmission and distribution lines between the monitored power supply equipment. The system uses node numbering to describe the busbar connection relationship, and uses a directed graph to represent the power flow direction of the transmission and distribution lines to build a complete network topology.

[0077] Monitoring equipment mapping: used to map the installation locations of various monitoring devices on the monitored power supply equipment and the collection scope of their monitoring data. The system uses the monitoring device as an information collection node, establishes an association with the monitored equipment, and clarifies the data coverage of each monitoring point.

[0078] The dynamic feature map constructed by the attribute extension model includes:

[0079] Power supply equipment feature mapping: used to construct the rated parameters of the monitored power supply equipment. For transformers, record the rated capacity, voltage level, short-circuit impedance, and connection group; for distribution lines, record the line length, conductor type, and unit impedance; for switchgear, record the rated current, rated capacity, and operating characteristics.

[0080] Operation status mapping: used to map the features collected by the monitoring device to the digital twin basic model. The system maps transient disturbance features, short-term fluctuation features, and steady-state features to the corresponding device nodes according to timestamps to form a dynamic mapping of the device's operation status.

[0081] Load feature mapping: Use the load data and historical operation data collected by the electrical equipment status monitoring device to build a load model. The system records the load capacity change pattern, power factor distribution characteristics, typical daily load curve, and establishes a load classification model.

[0082] The multi-scale analysis model includes:

[0083] Transient disturbance analysis model: After the transient disturbance characteristics are input into the model, the possible disturbance propagation path is first identified through the grid topology structure, and then the specific location of the disturbance source is calculated through the fault location algorithm in combination with the impedance parameters of each device, and finally the impact range of the disturbance is evaluated.

[0084] Short-term fluctuation analysis model: conduct time-series correlation analysis on short-term fluctuation characteristics and load curves, calculate the intensity level of the fluctuation source through the fluctuation propagation model, and use the impedance attenuation model to analyze the propagation and attenuation law of the fluctuation in the network to locate and evaluate the fluctuation source.

[0085] Steady-state characteristic analysis model: Compare the collected steady-state characteristics with historical operation data, identify the changing trends of system harmonic distortion and three-phase imbalance, determine the distribution locations of major pollution sources through propagation characteristic analysis, and evaluate their impact on the entire system.

[0086] This embodiment builds a complete digital twin model system, realizes comprehensive mapping of the physical properties, operating status and load characteristics of the distribution system, and provides reliable model support for power quality analysis.

[0087] As the fifth embodiment of the present invention

[0088] See also Figure 4 Based on Embodiments 1, 2, 3 and 4, this embodiment describes the three functional units of the application space layer in detail.

[0089] The power quality prediction unit preferably adopts a shallow neural network model with 2 hidden layers, and the number of hidden layer nodes is 16 and 8 respectively. The input layer receives the distortion source distribution characteristics (including the content of each harmonic, the phase angle of each harmonic) and the unbalanced propagation characteristics (including the negative sequence voltage amplitude, the negative sequence phase angle) output by the steady-state characteristic analysis model, as well as the daily load curve (including the 24-hour load rate and the power factor change curve) of the load characteristic mapping. Preferably, the output layer predicts the harmonic distortion rate, three-phase imbalance and power factor change trend in the distribution system within the next 24 hours. Preferably, the root mean square error is used as the loss function, and the Adam optimizer is used for training, and the training data is a 7-day rolling data set in the historical operation data.

[0090] The hierarchical early warning unit includes four early warning modules running in parallel:

[0091] The transient warning module achieves millisecond-level warning by comparing the received transient disturbance characteristics with the preset threshold. Preferably, the system sets the voltage sag depth threshold to 90% of the nominal voltage, the voltage swell amplitude threshold to 110% of the nominal voltage, and the interruption duration threshold to 20ms. When any indicator is detected to exceed the threshold, a warning message containing the fault type, occurrence time, duration and amplitude is generated.

[0092] The short-term warning module preferably uses a 10-minute sliding time window to process the received short-term fluctuation characteristics. The system sets the voltage fluctuation amplitude threshold to 3% of the nominal voltage and the duration threshold to 10 minutes. When the voltage fluctuation amplitude and duration exceed the threshold at the same time, a warning message including the start time, maximum fluctuation amplitude, and cumulative duration is generated.

[0093] The steady-state warning module preferably evaluates the received steady-state characteristics once every hour. Preferably, the system sets the total harmonic distortion rate limit to 5% and the voltage imbalance limit to 2%. When any indicator exceeds the limit, a warning message including the exceeding period, exceeding amplitude, and impact range is generated.

[0094] The trend warning module preferably processes the 24-hour prediction results output by the power quality prediction unit. Preferably, when the predicted harmonic distortion rate exceeds 5% or the three-phase imbalance exceeds 2%, warning information including the warning period, warning index and warning level is generated.

[0095] The governance decision-making unit includes two modes: emergency response and preventive deployment:

[0096] The emergency response mode adopts control measures for transient warning, short-term warning and steady-state warning signals respectively.

[0097] For transient warning signals, the transient disturbance analysis model is used to determine the location of the disturbance source based on the voltage-current relationship and the equipment impedance parameters, the voltage sensitivity analysis is used to determine the propagation path, and the optimal installation position of the dynamic voltage recovery device is determined according to the node voltage sensitivity in combination with the transient disturbance characteristics, and the required compensation capacity is calculated according to the voltage sag amplitude and the affected load capacity; As a specific embodiment of the present invention, when a certain distribution system executes the emergency disposal mode, the implementation process of the transient warning signal processing is as follows: When the voltage transient monitoring device detects that the A-phase voltage drops to 70% of the nominal value and the duration is 100ms, the transient disturbance analysis model first obtains the three-phase voltage and current at the fault point. According to the U=ZI relationship, combined with the impedance parameters of each device in the digital twin basic model (where the line impedance is 0.27+j0.35Ω / km and the transformer short-circuit impedance is 6.5%), the disturbance source location is calculated. The forward fault current method is used to determine that the disturbance source is located at the 10kV bus No. 2 outgoing line. Then, based on the voltage sensitivity analysis, the voltage sag amplitude ΔU of each node is calculated. sag,i , the formula is:

[0098]

[0099] Among them, ΔU sag,0 is the voltage sag amplitude at the disturbance source, Z i is the equivalent impedance from node i to the disturbance source, and Z0 is the equivalent impedance at the disturbance source. sag,i When it is greater than 10%, the node is determined to be affected by the disturbance. Calculations show that the voltage sag amplitudes of the three branch lines downstream of the outgoing line are 65%, 58% and 52% respectively, all exceeding the criteria, so it is determined to be within the disturbance range.

[0100] For short-term warning signals, a short-term fluctuation analysis model is used to determine the fluctuation source intensity and attenuation law based on the fluctuation propagation attenuation equation. Combined with the short-term fluctuation characteristics, the optimal installation position, required compensation capacity and number of devices of the static VAR generator are determined through the voltage fluctuation compensation equation. As another embodiment of the present invention, when processing short-term warning signals, when the system detects that the voltage fluctuation amplitude of bus No. 5 reaches 4% of the nominal value and the fluctuation frequency is 2Hz, the short-term fluctuation analysis model first establishes the fluctuation propagation equation:

[0101]

[0102] Where ΔU fluc,i is the voltage fluctuation amplitude at node i, ΔU fluc,s is the fluctuation amplitude at the fluctuation source, α is the attenuation coefficient (related to line parameters and load characteristics), d i is the distance from node i to the source of the fluctuation. By fitting the measured data, we get α to be 0.3 km -1, and then determine the metallurgical load where the fluctuation source is 2km downstream of the busbar. According to the compensation capacity calculation formula:

[0103]

[0104] Among them U n is the system nominal voltage, ΔU is the voltage fluctuation amplitude to be suppressed, k is the margin factor (taken as 1.5), and the required compensation capacity is calculated to be 385kvar. Considering the standard capacity of the equipment, two 200kvar static VAR generators are selected.

[0105] For steady-state warning signals, the steady-state characteristic analysis model is used to determine the distribution of distortion sources through harmonic flow calculation, and the unbalanced propagation characteristics are determined based on the three-phase unbalance calculation. Combined with the steady-state characteristics, the optimal installation position, required compensation capacity and number of devices of the active filter are determined by using the comprehensive compensation capacity of harmonics and unbalance. As another embodiment of the present invention, when executing the steady-state warning signal processing, the system monitors that the total harmonic distortion rate on the low-voltage side of the No. 3 distribution transformer reaches 6.2%, mainly the 5th harmonic (5.1%), and the three-phase imbalance is 2.5%. First, the compensation capacity requirement of each harmonic is determined based on the harmonic flow calculation, and the formula is:

[0106]

[0107] Among them, S h is the compensation capacity of the hth harmonic, U h is the effective value of the harmonic voltage of order h, I h is the effective value of the harmonic current of order h. After calculation, the compensation capacity requirements of the 5th / 7th / 11th harmonics are 95kVA, 45kVA and 25kVA respectively. At the same time, the formula for the three-phase unbalanced compensation capacity requirement is:

[0108] S u =3×U n ×I 2

[0109] Among them, U n is the system nominal voltage, I 2 is the negative sequence current, S u is the three-phase unbalanced compensation capacity. The unbalanced compensation capacity requirement is calculated to be 35kVA. Considering a 30% margin, the formula for the total compensation capacity requirement is:

[0110]

[0111] Where S is the total compensation capacity demand, k sis the margin factor. Based on the calculation results of harmonic compensation capacity (including 5 / 7 / 11 harmonic compensation capacity of 95 / 45 / 25kVA respectively) and unbalance compensation capacity (35kVA), the calculation formula of harmonic and unbalanced comprehensive compensation capacity is adopted, and the 30% margin factor is considered. The total compensation capacity requirement is calculated to be about 147kVA. According to the standard capacity gear of the active filter, the 150kVA specification is selected to ensure that the comprehensive compensation requirements of the system are met.

[0112] The preventive deployment mode is configured and planned based on the prediction results. Based on the harmonic distortion rate and three-phase imbalance trend output by the power quality prediction unit, harmonic superposition calculation is used to identify potential quality problem areas; combined with the distortion source distribution and unbalanced propagation characteristics of the steady-state characteristic analysis model, the installation location and capacity requirements of the compensation device in the problem area are determined through the optimal configuration of the compensation effect; a long-term configuration plan including short-term measures and long-term planning is formulated. As another embodiment of the present invention, in the preventive deployment mode, the power quality prediction unit uses a shallow neural network model with 2 hidden layers (the number of hidden layer nodes is 16 and 8 respectively) to predict the future harmonic level in the No. 4 distribution station area. The model input includes: the distortion source distribution characteristics and unbalanced propagation characteristics output by the steady-state characteristic analysis model, and the daily load curve mapped by the load characteristics, and the output is the 24-hour harmonic prediction result. The prediction shows that the content of each harmonic will reach: 4.2% for the 5th, 2.8% for the 7th, 1.9% for the 11th, and 1.6% for the 13th, with a total distortion rate of 5.8%. The required compensation capacity is calculated based on the principle of harmonic superposition, and the formula is:

[0113]

[0114] Among them, S is the total compensation capacity demand, S n is the compensation capacity of the nth harmonic, n is the harmonic number, N is the highest harmonic number considered, K s is the margin coefficient. Based on the harmonic prediction results (including 5 / 7 / 11 / 13 harmonic contents of 4.2% / 2.8% / 1.9% / 1.6% respectively), the required capacity in the near term is calculated to be 185kVA, and the 200kVA specification is selected. Considering the harmonic characteristics of the new loads in the regional development plan, 3×100kVA distributed compensation devices need to be added in the long term. The specific installation location of each device is determined by optimizing the objective function:

[0115]

[0116] Among them, F is the optimization objective function, THD i is the harmonic distortion rate of node i, L i is the load capacity of node i, i is the node number, n is the total number of system nodes, and the constraints include the capacity limit of the compensation equipment and the installation condition limit.

[0117] Through the coordinated cooperation of the three functional units, this embodiment achieves accurate prediction, graded warning and precise management of power quality problems, providing effective support for ensuring the power supply quality of the distribution system.

[0118] As the sixth embodiment of the present invention

[0119] This embodiment integrates Embodiments 1 to 5 to form a complete power quality prediction and management platform.

[0120] The platform adopts a hierarchical architecture, which is constructed from the physical space layer to the application space layer. The monitoring devices deployed in the physical space layer collect raw data, the data space layer processes it according to multiple time scales, the model space layer builds a digital twin model system, and the application space layer realizes the prediction, early warning and governance decision-making functions.

[0121] Its workflow is:

[0122] Various monitoring devices in the physical space layer collect data on voltage transients, voltage fluctuations, harmonics, three-phase imbalance, and equipment operating status in real time. The monitoring devices use a unified clock synchronization mechanism to ensure the timing consistency of the data.

[0123] After receiving the monitoring data, the data space layer processes transient disturbances, short-term fluctuations and steady-state characteristic data according to three sampling frequencies: above 10kHz, minute level and hour level. A sampling strategy combining event triggering and fixed period is adopted to form a characteristic data set.

[0124] The model space layer establishes physical entity mapping based on the digital twin basic model, constructs dynamic feature mapping through the attribute extension model, and uses the multi-scale analysis model to realize system dynamic analysis. The three models work together to realize the mapping from physical space to digital space.

[0125] The application space layer performs algorithm processing based on the model analysis results. Among them, the power quality prediction unit predicts the change trend of harmonic distortion rate, three-phase imbalance and power factor through a shallow neural network model, the hierarchical early warning unit realizes multi-level early warning through four early warning modules: transient, short-term, steady state and trend, and the governance decision-making unit formulates compensation device configuration plans through two modes: emergency response and preventive deployment.

[0126] This embodiment realizes closed-loop control of power quality monitoring, analysis, prediction, early warning and management through the organic integration of functions at various layers, and provides a systematic solution for improving the power supply quality of the distribution system.

[0127] Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0128] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A power quality prediction and management platform based on digital twin technology, characterized in that: It includes physical space layer, data space layer, model space layer and application space layer; The physical space layer is used to deploy power quality monitoring devices and electrical equipment status monitoring devices to collect voltage transients, voltage fluctuations, harmonics, three-phase imbalance and equipment operating status data; The data space layer performs data collection and preprocessing according to multiple time scales, including the collection and preprocessing of transient disturbance data, short-term fluctuation data and steady-state characteristic data; The model space layer includes a digital twin basic model, an attribute extension model and a multi-scale analysis model, wherein the digital twin basic model is used to construct a physical entity mapping of the distribution system, the attribute extension model is used to construct a dynamic feature mapping based on the digital twin basic model, and the multi-scale analysis model is used to realize dynamic analysis of the distribution system based on the data collected by the data space layer; The application space layer includes a power quality prediction unit, a hierarchical early warning unit and a governance decision unit, wherein the power quality prediction unit is used to predict the changing trend of harmonic distortion rate and three-phase imbalance, the hierarchical early warning unit is used to perform real-time, short-term and trend early warnings, and the governance decision unit is used to formulate emergency compensation and preventive deployment plans.

2. The power quality prediction and management platform according to claim 1 is characterized in that: The physical space layer includes: The power quality monitoring device includes a voltage transient monitoring device, a voltage fluctuation monitoring device, a harmonic monitoring device and a three-phase unbalance monitoring device, wherein the voltage transient monitoring device is used to collect voltage sag, voltage swell and voltage interruption data, the voltage fluctuation monitoring device is used to collect voltage fluctuation and voltage flicker data, the harmonic monitoring device is used to collect harmonic distortion data, and the three-phase unbalance monitoring device is used to collect three-phase voltage unbalance data; The electrical equipment status monitoring device is used to collect the transformer load rate, the distribution line load current and the operating status of the switchgear.

3. The power quality prediction and management platform according to claim 1, characterized in that: The multi-time scale data acquisition and preprocessing includes: Transient disturbance data collection, using a sampling frequency of more than 10kHz, is used to monitor voltage sags, swells and interruption events. High-frequency sampling is performed through the voltage transient monitoring device, and the collected data is decomposed by wavelet transform to extract transient disturbance characteristics, including voltage sag amplitude, duration and waveform distortion; Short-term fluctuation data collection, using minute-level sampling frequency, is used for voltage fluctuation and flicker monitoring. Periodic sampling is performed through the voltage fluctuation monitoring device. After variance analysis of the collected data, short-term fluctuation characteristics are extracted, including fluctuation amplitude and fluctuation frequency. Steady-state characteristic data collection, using hourly sampling frequency, is used for monitoring harmonics, three-phase imbalance and load changes. Regular sampling is performed through the harmonic monitoring device, three-phase imbalance monitoring device and electrical equipment status monitoring device. After the collected data is analyzed, steady-state characteristics are extracted, including harmonic spectrum, imbalance and load characteristics; The multi-time scale data is spatially synchronized with the clocks of each monitoring device through time synchronization technology to achieve synchronous collection, and is temporally preprocessed according to a sampling strategy that combines an event triggering method with a fixed period method to form a feature data set for different analysis targets.

4. The power quality prediction and management platform according to claim 1, characterized in that: The physical entity mapping constructed by the digital twin basic model includes: A monitored power supply equipment mapping, used to map the physical connection relationship between transformers, distribution lines and switchgear in the physical space layer; Grid topology mapping, used to map the bus node connection relationship and the direction of transmission and distribution lines between the monitored power supply equipment; Monitoring equipment mapping is used to map the installation locations of the various monitoring devices on the monitored power supply equipment and the collection scope of their monitoring data.

5. The power quality prediction and management platform according to claim 1, characterized in that: The dynamic feature map constructed by the attribute extension model includes: Power supply equipment feature mapping, used to construct the rated parameters of the monitored power supply equipment, including transformer rated capacity and short-circuit impedance, distribution line impedance parameters, and rated capacity of switchgear; Operation state mapping, used to map the transient disturbance characteristics, short-term fluctuation characteristics and steady-state characteristics collected by the monitoring device onto the digital twin basic model; Load characteristic mapping uses the load data and historical operation data collected by the electrical equipment status monitoring device to construct the capacity, power factor and daily load curve of the load carried by the transformer and distribution line.

6. The power quality prediction and management platform according to claim 1, characterized in that: The multi-scale analysis model specifically includes: A transient disturbance analysis model, inputting the transient disturbance characteristics into the model, calculating the disturbance propagation path through the structural relationship of the power grid topology mapping, and calculating the disturbance source location in combination with the impedance parameters of the power supply equipment characteristic mapping; A short-term fluctuation analysis model is used to correlate the short-term fluctuation characteristics with the daily load curve characteristics mapped by the load characteristics, and calculate the intensity and attenuation law of the fluctuation source; The steady-state characteristic analysis model compares and analyzes the steady-state characteristic with the steady-state characteristic of the operating state mapping, and calculates the distortion source distribution and the unbalanced propagation characteristics.

7. The power quality prediction and management platform according to claim 1, characterized in that: The application space layer includes a power quality prediction unit, a hierarchical early warning unit and a governance decision unit; Among them, the power quality prediction unit predicts the changing trends of harmonic distortion rate, three-phase imbalance and power factor in the distribution system through a shallow neural network model, based on the distortion source distribution and unbalanced propagation characteristics output by the steady-state characteristic analysis model and the daily load curve mapped by the load characteristics.

8. The power quality prediction and management platform according to claim 7, characterized in that: The hierarchical early warning unit includes four modules: The transient warning module receives transient disturbance characteristics and triggers millisecond-level warnings when it detects that the transient disturbance exceeds the preset threshold; The short-term warning module receives short-term fluctuation characteristics and triggers a warning when it detects that the voltage fluctuation exceeds a preset threshold and lasts for more than a preset time; The steady-state warning module receives steady-state characteristics and triggers a warning when it detects that the harmonic content or imbalance exceeds the standard limit; The trend warning module receives the quality degradation trend prediction of the power quality prediction unit and triggers a warning when the predicted value of the power quality indicator exceeds the standard limit.

9. The power quality prediction and management platform according to claim 7, characterized in that: The governance decision-making unit includes two modes: emergency response and preventive deployment: The emergency response mode, for transient warning, short-term warning and steady-state warning signals, respectively uses the disturbance source location and propagation path determined by the transient disturbance analysis model in combination with the transient disturbance characteristics to determine the optimal installation location and required compensation capacity of the dynamic voltage recovery device, uses the fluctuation source intensity and attenuation law determined by the short-term fluctuation analysis model in combination with the short-term fluctuation characteristics to determine the optimal installation location, required compensation capacity and number of devices of the static VAR generator, and uses the distortion source distribution and unbalanced propagation characteristics determined by the steady-state characteristic analysis model in combination with the steady-state characteristics to determine the optimal installation location, required compensation capacity and number of devices of the active filter; the preventive deployment mode, based on the harmonic distortion rate and three-phase imbalance trend predicted by the power quality prediction unit, identifies potential quality problem areas, determines the installation location and capacity requirements of the compensation device in the problem area in combination with the distortion source distribution and unbalanced propagation characteristics of the steady-state characteristic analysis model, and formulates a long-term configuration planning plan.

Images