Power grid data fusion method, system and equipment

Abstract

The invention provides a power grid data fusion method, system and device. The power grid data fusion method comprises the following steps: acquiring power grid data acquired by SCADA, WAMS and RPMS to generate a multi-source data set; a fusion analysis result is obtained through the multi-source data set, the fusion analysis result comprises comprehensive operation indexes and abnormal information, the comprehensive operation indexes are obtained through principal component analysis processing, and the abnormal information is obtained through abnormal recognition processing; and providing the fusion analysis result to the application terminal. The fault processing efficiency of a power grid enterprise and the user experience of the power grid service can be improved while early recognition and accurate positioning of the power grid abnormity are realized.

Description

Technical Field

[0001] This disclosure relates to the field of multi-sensor fusion technology, and more specifically, to a power grid data fusion method, system, and device. Background Technology

[0002] Smart grid operation and management involves key business processes across the entire power generation and consumption chain, including load forecasting, operation monitoring, equipment maintenance, power trading, and customer services. To optimize the allocation of grid resources and processes and improve operational efficiency and service quality, grid companies are increasingly reliant on integrated cyber-physical systems and advanced data analytics technologies. In this process, operational data fusion serves as a key supporting technology, aiming to effectively integrate, analyze, and mine multi-source, heterogeneous operational data from both internal and external sources. This optimizes grid operation management and service processes, thereby improving grid resource utilization while reducing grid operating costs.

[0003] However, when processing multi-source operational data from power grid companies, related technologies generally suffer from information silos. For example, systems such as Supervisory Control and Data Acquisition (SCADA), Wide Area Measurement System (WAMS), and Relay Protection Fault Information Management System (RPMS) operate independently, with varying data formats, communication protocols, and storage methods. This leads to low efficiency in data integration and circulation. This decentralized data management approach prevents the full utilization of data value, severely restricting the comprehensiveness, real-time performance, and consistency of data in decision support, resulting in low data value utilization and limiting the comprehensiveness and real-time performance of data during decision-making.

[0004] Furthermore, the data analysis of related technologies is mostly in a static and passive processing mode, lacking sufficient intelligence and initiative, making it difficult to adapt to the dynamic changes in the power grid business environment, such as load fluctuations, equipment status changes, and real-time changes in user service needs. Because effective data fusion cannot achieve early identification and accurate location of service anomalies, problems are often only noticed after they have escalated, resulting in delayed service response, impacting fault handling efficiency and the end-user service experience, thereby hindering the improvement of the overall service quality of power grid companies.

[0005] Therefore, a power grid data fusion solution is needed to achieve efficient integration, intelligent analysis, and proactive response of multi-source heterogeneous data, thereby enabling early identification and accurate location of service anomalies, improving fault handling efficiency and user experience. Summary of the Invention

[0006] To address the aforementioned technical problems, embodiments of this disclosure provide a power grid data fusion method, system, and device.

[0007] According to one aspect of this disclosure, a power grid data fusion method is provided, the method comprising: Acquire power grid data collected by SCADA, WAMS, and RPMS to generate a multi-source dataset; The multi-source dataset is used to obtain fusion analysis results, which include comprehensive operating indicators and anomaly information. The comprehensive operating indicators are obtained through principal component analysis, and the anomaly information is obtained through anomaly identification. The fusion analysis results are provided to the application terminal.

[0008] In some embodiments of the first aspect of this disclosure, the step of acquiring power grid data collected by the SCADA system, WAMS system, and RPMS system to generate a multi-source dataset includes: Sensors deployed at each node of the power grid collect power grid operation data for the corresponding node. SCADA acquires power grid operation data collected by sensors at various nodes of the power grid to generate second-level steady-state data, which includes both digital and analog signals. WAMS acquires power grid operation data collected by sensors at each node of the power grid to generate millisecond-level synchronous phasor data. The millisecond-level synchronous phasor data includes the amplitude and phase angle of the voltage fundamental phasor and the amplitude and phase angle of the current fundamental phasor. RPMS acquires power grid operation data collected by sensors at each node of the power grid to generate fault event data. The fault event data includes protection trip commands, switch quantity change sequences, and instantaneous waveforms of current and voltage before and after the fault. The second-level steady-state data, the millisecond-level synchronization phasor data, and the fault event data are aligned, labeled, and structured and stored based on a preset standardized timestamp to obtain the multi-source dataset.

[0009] In some embodiments of the first aspect of this disclosure, the method further includes: preprocessing the multi-source dataset before obtaining the fusion analysis result using the multi-source dataset, the preprocessing including one or more of the following: removing duplicate data, filling in missing data, normalization processing, and standardization processing.

[0010] In some embodiments of the first aspect of this disclosure, the comprehensive operating index is obtained by: standardizing the multi-source dataset to eliminate dimensions and obtaining an original data matrix; calculating the covariance matrix using the original data matrix; determining the number of principal components by solving for eigenvalues ​​and eigenvectors based on the covariance matrix; calculating the contribution rate of each principal component and the cumulative contribution rate of multiple principal components; and constructing a principal component score matrix, wherein the principal component score matrix is ​​the comprehensive operating index.

[0011] In some embodiments of the first aspect of this disclosure, the anomaly information and the comprehensive operating indicators are obtained through a pre-trained Transformer model, which is used to perform the principal component analysis processing and the anomaly identification processing in parallel.

[0012] In some embodiments of the first aspect of this disclosure, after providing the fusion analysis results to the application terminal, the method further includes one or more of the following: Frequency stability assessment data is obtained based on the fusion analysis results; power supply adjustment information is generated based on the frequency stability assessment data and the load forecast data obtained from the power supply side application terminal, the power supply adjustment information includes key information on the adjustment characteristics of various generator sets, output suggestion information and response timing; Frequency information is obtained based on the fusion analysis results; based on the frequency information and the line power flow information, load forecast information and new energy output information obtained by the second application terminal, control strategy suggestion information is generated, the control strategy suggestion information includes energy storage site recommendation information, action timing information, power range information and duration information; Based on the fusion analysis results, the historical operation and maintenance data of the specified node is processed to remove impurities in order to obtain the operation and maintenance data quality information of the specified node.

[0013] According to a second aspect of this disclosure, a power grid data fusion system is provided, the power grid data fusion system comprising: The data acquisition unit is used to acquire power grid data collected by SCADA, WAMS and RPMS to generate multi-source datasets; The fusion analysis unit is used to obtain fusion analysis results using the multi-source dataset. The fusion analysis results include comprehensive operating indicators and abnormal information. The comprehensive operating indicators are obtained through principal component analysis, and the abnormal information is obtained through information identification and extraction. The result providing unit is used to provide the fusion analysis results to the application terminal.

[0014] In some embodiments of the second aspect of this disclosure, the data acquisition unit is specifically used to: align, label, and structure the second-level steady-state data generated by the SCADA acquiring power grid operation data collected by sensors at each node of the power grid, the millisecond-level synchronization phasor data generated by the WAMS acquiring power grid operation data collected by sensors at each node of the power grid, and the fault event data generated by the RPMS acquiring power grid operation data collected by sensors at each node of the power grid, based on a preset standardized timestamp, to obtain the multi-source dataset.

[0015] In some embodiments of the second aspect of this disclosure, the power grid data fusion system further includes: a data preprocessing unit for preprocessing the multi-source dataset, wherein the preprocessing includes one or more of the following: removing duplicate data, filling in missing data, normalization processing, and standardization processing.

[0016] According to a third aspect of this disclosure, an electronic device is provided, the electronic device including a processor and a memory, the memory storing a computer program that, when executed by the processor, causes the processor to perform the methods described above.

[0017] This disclosure utilizes data from multiple systems such as SCADA, WAMS, and RPMS to form a multi-source dataset, ensuring the comprehensiveness and accuracy of data acquisition and breaking down information silos at their source. By performing feature extraction and principal component analysis on the multi-source dataset, deep fusion of heterogeneous data is achieved to obtain fusion analysis results. These results simultaneously include anomaly information and comprehensive operational indicators, significantly improving the value density and usability of power grid data. Furthermore, it enhances the ability to perceive operational status and identify service anomalies early, enabling flexible responses to dynamic changes during operation, such as load fluctuations or uncertainties brought about by renewable energy integration. Finally, the fusion analysis results are provided to user terminals such as the power generation side, grid side, and load side, providing data support and decision support to improve the execution efficiency of power generation planning, energy storage scheduling, and demand response, while also achieving cross-departmental resource optimization and process collaboration. Therefore, this disclosure improves the execution efficiency of multiple sides, such as the power generation side, grid side, and load side, through power grid data fusion, achieving early identification and accurate location of power grid anomalies while improving the fault handling efficiency of power grid companies and the user experience of power grid services. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the power grid data fusion method provided in this embodiment of the disclosure; Figure 2 This is a schematic diagram of the multi-source dataset generation process involved in the embodiments of this disclosure; Figure 3 This is a schematic diagram of the process for obtaining comprehensive operating indicators involved in the embodiments of this disclosure; Figure 4 This is a schematic diagram of the structure of the power grid data fusion system provided in the embodiments of this disclosure; Figure 5 This is a schematic diagram of the architecture of the power grid data fusion system provided in the embodiments of this disclosure; Figure 6 This is a schematic structural block diagram of an electronic device provided in an embodiment of this disclosure. Detailed Implementation

[0019] Hereinafter, exemplary embodiments according to the present disclosure will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present disclosure, and not all embodiments of the present disclosure, and it should be understood that the present disclosure is not limited to the exemplary embodiments described herein.

[0020] Figure 1 A schematic flowchart of the power grid data fusion method provided in an embodiment of this disclosure is shown. See also... Figure 1 The power grid data fusion method of this disclosure includes the following steps: Step 101: Acquire power grid data collected by SCADA, WAMS, and RPMS to generate a multi-source dataset; Step 102: Obtain the fusion analysis results using multi-source datasets. The fusion analysis results include comprehensive operating indicators and anomaly information. The comprehensive operating indicators are obtained through principal component analysis, and the anomaly information is obtained through anomaly identification. Step 103: Provide the fusion analysis results to the application terminal.

[0021] The embodiments disclosed herein can achieve efficient integration, intelligent analysis, and proactive response of multi-source heterogeneous data, thereby enabling early identification and accurate location of power grid anomalies, while improving the fault handling efficiency of power grid enterprises and the user experience of power grid services.

[0022] Figure 2 A schematic diagram illustrating an exemplary implementation of step 101 is shown. See also... Figure 2 An exemplary implementation of step 101, which generates a multi-source dataset, may include the following steps 201 to 203: Step 201: Sensors deployed at each node of the power grid collect power grid operation data at the corresponding nodes of the power grid. Sensors are applied at various nodes in a power system to monitor the operating status of electrical equipment at each node in real time. Sensors enable continuous monitoring of the operating status of critical equipment in the power grid.

[0023] Sensors can include various types of sensors deployed on power equipment. For example, sensors may include, but are not limited to: oil temperature monitoring units installed on transformers, micro-meteorological and video monitoring devices deployed on transmission lines, and partial discharge detection sensors integrated in switchgear.

[0024] Sensors are deployed at various key nodes in the power system, such as power generation, transmission, transformation, and distribution, to collect real-time operating status data of various power equipment at the corresponding nodes, extract data source information and acquisition frequency, generate power grid operation data including equipment status, data source and acquisition frequency, and provide it to SCADA, WAMS and RPMS.

[0025] The data acquisition frequency of sensors can be flexibly configured according to the importance, parameter characteristics, and safety level of power equipment at the corresponding nodes of the power grid. For example, critical current and voltage quantities used for protection tripping can be acquired at millisecond-level high-speed acquisition, while transformer winding temperature data used to assess the health status of equipment can be updated at the second or minute level. Through this step, a network-wide, time-series-defined equipment status dataset can be obtained and merged to generate standardized power grid operation data. This power grid operation data includes power equipment status data and information such as its data source and acquisition frequency.

[0026] Step 202: SCADA, WAMS and RPMS acquire power grid operation data collected by sensors at each node of the power grid to generate second-level steady-state data, millisecond-level synchronization phasor data and fault event data; SCADA can acquire power grid operation data collected by sensors at various nodes of the power grid to generate second-level steady-state data, which includes both digital and analog signals.

[0027] Specifically, SCADA collects the aforementioned power grid operation data through remote terminal units (RTUs) and intelligent electronic devices (IEDs) deployed in power plants and substations. The master station uses a combination of polling and event-driven methods to obtain information such as the data source and collection frequency. SCADA can obtain second-level steady-state data from the power grid operation data provided by various sensors. This second-level steady-state data includes both digital and analog quantities. Digital quantities include, but are not limited to, circuit breaker positions, disconnector status, and relay protection signals. Analog quantities include, but are not limited to, line current, bus voltage, and active / reactive power. The characteristics of second-level steady-state data are wide coverage, high standardization, and time-stamped data. Among them, digital quantities include switch position information, which is an event record of changes in digital quantities. Switch position information is reported immediately (millisecond level) in an event-driven manner. Analog quantities can be collected by polling at preset intervals (second to minute level), which is suitable for monitoring the steady-state operation of the power grid and provides basic and continuous operating parameters. For example, in a 110kV substation, SCADA polls and collects the current, voltage, and power measurements of the entire station every 2 seconds. When a fault in a feeder causes the circuit breaker to trip, the switch position change signal is immediately reported, and the switch quantity such as the change in the position of the circuit breaker and the corresponding current value is obtained.

[0028] WAMS can acquire power grid operation data collected by sensors at various nodes of the power grid to generate millisecond-level synchronous phasor data. The millisecond-level synchronous phasor data includes the amplitude and phase angle of the voltage fundamental phasor and the amplitude and phase angle of the current fundamental phasor.

[0029] Specifically, WAMS performs measurements using phasor measurement units (PMUs) based on GPS synchronization time scales, and the data is transmitted to a phasor data concentrator via a communication network. The millisecond-level synchronized phasor data generated by WAMS includes, but is not limited to, the amplitude and phase angle of the fundamental voltage phasor and the amplitude and phase angle of the fundamental current phasor. The characteristics of millisecond-level synchronized phasor data are high network-wide time synchronization and fast sampling rates (typically tens to hundreds of frames per second), enabling precise capture of electromechanical transient processes and dynamic behaviors of the power grid (such as low-frequency oscillations). It is suitable for power grid dynamic characteristic analysis and stability assessment, providing high-precision synchronized dynamic data. For example, to monitor low-frequency oscillations in a regional power grid, WAMS can acquire PMU data from multiple 500kV substations at a rate of 50 frames per second, including the amplitude and phase angle of the voltage phasor, thereby recording frequency fluctuations and disturbance propagation trajectories.

[0030] RPMS can acquire power grid operation data collected by sensors at each node of the power grid to generate fault event data. Fault event data includes protection trip commands, switch quantity change sequences, and instantaneous waveforms of current and voltage before and after the fault. Specifically, RPMS can be triggered to collect data during power grid faults via intelligent protection devices and fault recorders, operating on an event-driven mechanism. The fault event data generated by RPMS includes, but is not limited to, protection trip commands, switch change sequences, and instantaneous current and voltage waveforms before and after the fault. The fault event data is characterized by its event-driven nature, high recording accuracy, and large data depth (i.e., millisecond-level time resolution), enabling it to completely capture the entire fault waveform process. It is suitable for fault cause tracing, protection action behavior evaluation, and accident reversal, providing detailed diagnostic evidence. For example, when a transient fault occurs on a 220kV line, RPMS can obtain the protection device's trip report, fault location results, and current and voltage waveform data from 200 to 1000 milliseconds before and after the fault.

[0031] Step 203: Align, label, and structure the second-level steady-state data, millisecond-level synchronization phasor data, and fault event data based on the preset standardized timestamps to obtain a multi-source dataset.

[0032] To achieve unified perception and integrated analysis of the power grid's operating status, the data from each independent system (i.e., SCADA, WAMS, and RPMS) need to be integrated into a multi-source dataset. This dataset consists of the second-level steady-state data provided by SCADA, the millisecond-level synchronization phasor data provided by WAMS, and the fault event data from the RPMS system mentioned in the previous steps.

[0033] The integration process is not a simple accumulation, but rather a process using a data engine to align, label, and structure data from different sources (SCADA, WAMS, RPMS) and with different sampling rates (i.e., second-level steady-state data, millisecond-level synchronous phasor data, and fault event data) based on a unified standardized timestamp, thereby forming a multi-source dataset. For example, the average voltage per minute for a bus in SCADA, the instantaneous voltage phasor values ​​of that bus every 20 milliseconds in the same time period in WAMS, and any protection action events recorded by RPMS that occurred during that time period can be correlated and integrated on a unified time axis. The resulting multi-source dataset is a temporally synchronized and semantically standardized panoramic dataset of the power grid, containing information on slow-changing steady-state, fast-changing dynamic, and sudden events.

[0034] As described above, this embodiment utilizes multi-source heterogeneous systems such as SCADA, WAMS, and RPMS to acquire various types of real-time data from each node in the power grid, and integrates these real-time data into a multi-source dataset for subsequent principal component analysis and anomaly identification processing.

[0035] Furthermore, before obtaining the fusion analysis results using the multi-source dataset in step 102, the method of this embodiment may further include: preprocessing the multi-source dataset. The preprocessing of the multi-source dataset may include, but is not limited to, one or more of the following: removing duplicate data, filling in missing data, normalization, and standardization. Preprocessing can obtain a complete multi-source dataset and ensure that the data quality and consistency of the multi-source dataset meet the requirements of the relevant processing in step 102.

[0036] Specifically, the preprocessing of multi-source datasets can be flexibly selected based on the severity and distribution characteristics of missing data. The preprocessing standards and relevant thresholds can be predefined by relevant data quality specifications. For example, for sequences with a single data point missing rate of less than 5%, the mean of other normal data at the same time point can be used for imputation; for sequences with a missing rate exceeding 20%, multiple imputation can be initiated, constructing five imputation datasets and merging them for analysis to assess uncertainty. Another example is when processing substation temperature data where a sensor loses three consecutive data points (minute-level) due to communication interruption; the mean of the preceding and following time points can be automatically used for imputation. Yet another example is when a PMU channel experiences a large loss of one hour's data due to a fault; its regression model with related electrical quantities (such as power and voltage) is invoked to generate five possible reasonable data sequences, and their average value is taken as the final imputation value, followed by normalization within the range [0, 1].

[0037] Before obtaining fusion analysis results from multi-source datasets, preprocessing the multi-source datasets can effectively remove noise from the original data, eliminate data missing and inconsistency issues, and generate a clean, complete, and scale-uniform standard dataset, thereby improving the accuracy and efficiency of subsequent processing.

[0038] Principal component analysis can transform a set of potentially related variables (i.e., features) into a set of linearly unrelated principal components. Figure 3 A schematic diagram illustrating an exemplary process for obtaining comprehensive operational metrics in step 102 is shown. See also... Figure 3 The comprehensive operating indicators can be obtained through the following steps 301 to 303: Step 301: Standardize the multi-source dataset to eliminate dimensions and obtain the original data matrix; Specifically, the multi-source datasets are standardized to eliminate the influence of units.

[0039] Transform multi-source datasets into a single dataset with... Features and The original data matrix of each sample is Each element in the original data matrix satisfies the following equation: , in, The original data matrix obtained by standardization processing The elements in Indicates the first The mean of each feature, It is the first The standard deviation of each feature , .

[0040] For example, in multi-source datasets, each point in the second-level steady-state data from SCADA, such as circuit breaker position, disconnector status, relay protection signals, line current, bus voltage, and active / reactive power, can be considered a feature. Similarly, in multi-source datasets, the amplitude, phase angle, amplitude, and phase angle of the fundamental voltage phasor, and the fundamental current phasor, respectively, in the millisecond-level synchronous phasor data from WAMS, can each be considered a feature. Furthermore, each of the following in multi-source datasets, such as protection trip commands, switch quantity change sequences, and instantaneous current and voltage waveforms before and after a fault, can be considered a feature from fault event data from RPMS. In addition, other features can be included; for example, temperature, humidity, and wind speed can be considered as features affecting load, and total regional electricity consumption can be considered as a demand-side feature.

[0041] Step 302: Calculate the covariance matrix using the original data matrix. The covariance matrix describes the linear relationship between the features. Specifically, the covariance matrix can be calculated using the following formula.

[0042] , in, The covariance matrix of the original data matrix. This represents the original data matrix.

[0043] Step 303: Determine the number of principal components by solving for eigenvalues ​​and eigenvectors based on the covariance matrix, calculate the contribution rate of each principal component and the cumulative contribution rate of multiple principal components, and construct the principal component score matrix. The principal component score matrix is ​​the comprehensive operating index.

[0044] Specifically, the number of principal components can be determined by solving for the eigenvalues ​​and eigenvectors, the contribution rate of each principal component can be calculated, the cumulative contribution rate of multiple principal components can be calculated, and the principal component score matrix can be constructed.

[0045] , , , , in, This represents the eigenvalue decomposition of the covariance matrix of the original data matrix. Indicates the first 1 eigenvalue, Indicates the first The unit eigenvector corresponding to each eigenvalue. . For the first The contribution rate of each principal component For the front The cumulative contribution rate of each principal component This indicates the number of principal components, which can be selected based on the magnitude of the eigenvalues. Each principal component can be determined based on the contribution rate of an individual principal component and the cumulative contribution rate. For example, you can choose This causes the cumulative explanatory rate to exceed a predetermined threshold (e.g., 80%, 95%, etc.). for Before each sample Principal component score matrix of principal components. For the front The matrix formed by the eigenvalues ​​and their corresponding eigenvectors This represents the original data matrix.

[0046] The following example illustrates the principal component analysis process for multi-source datasets.

[0047] Suppose that data on current I, voltage V, and power P are collected at five different time points at a node in the power grid, and the corresponding raw data... Medium sample size 5. Number of features The value is 3.

[0048] Raw data It can be expressed as the following formula: , The original data matrix obtained after standardization It can be represented as: , The calculated covariance matrix is ​​as follows: , Based on the covariance matrix, the eigenvalues ​​and eigenvectors are obtained, and the contribution rate, cumulative contribution rate, and principal component score matrix are calculated as follows: , , , , , As can be seen from the above, principal component analysis can successfully transform the three highly correlated operating variables (i.e., current, voltage, and power) into a comprehensive operating index. This index captures 98.9% of the information in the original data. The highest score is 2.515, which corresponds to the state in which current, voltage, and power are all at their highest in the original data; the lowest score is -2.738, which corresponds to the state in which all values ​​are at their lowest.

[0049] In step 102, the abnormal information may include, but is not limited to, numerical values ​​related to fault conditions, abnormal operation, etc., from the multi-source dataset. For example, the abnormal information may include the abnormal current value generated when a short circuit occurs, which is identified and extracted from the multi-source dataset.

[0050] Furthermore, anomaly information and comprehensive operational metrics can be obtained through a pre-trained Transformer model, which can be used to perform principal component analysis and anomaly identification.

[0051] Therefore, the Transformer model, which is capable of processing complex spatiotemporal data, can be used to deeply mine and analyze multi-source datasets to accurately extract abnormal information and generate comprehensive operational indicators, thereby achieving advanced perception and intelligent decision-making regarding the power grid's operational status.

[0052] In practical applications, the training and application process of the Transformer model varies depending on the business objectives. Different model configuration information can be generated for different business scenarios. Using this model configuration information and pre-collected sample data, a Transformer model capable of simultaneously extracting anomaly information and generating comprehensive performance metrics can be trained.

[0053] Based on a constructed multi-source complete dataset, the Transformer model is used for deep feature extraction and multi-indicator fusion analysis to achieve intelligent perception of the power grid's operating status. This model can not only identify key operating characteristics but also support fault analysis and processing as well as real-time operational status early warning. Ultimately, it outputs fusion analysis results containing anomaly information and comprehensive operating indicators, providing data support for precise monitoring and decision-making by power grid companies.

[0054] As described above, the embodiments of this disclosure can enhance the ability to perceive the operational status of services and identify service anomalies early by using a dynamic data processing and fusion mechanism based on artificial intelligence. This enables flexible responses to dynamic changes during operation, such as load fluctuations or uncertainties brought about by renewable energy access.

[0055] After step 103, the method of this disclosure embodiment may further include one or more of the following: 1) Frequency stability assessment data is obtained based on the fusion analysis results; power supply adjustment information is generated based on the frequency stability assessment data and the load forecast data obtained from the power supply side application terminal. The power supply adjustment information includes key information on the adjustment characteristics of various generator sets, output recommendation information and response timing. 2) Frequency information is obtained based on the fusion analysis results; based on the frequency information and the line power flow information, load forecast information and new energy output information obtained from the grid-side application terminal, control strategy suggestion information is generated, including energy storage site recommendation information, action timing information, power range information and duration information; 3) Based on the fusion analysis results, the historical operation and maintenance data of the specified node is processed to remove impurities in order to obtain the operation and maintenance data quality information of the specified node.

[0056] Specifically, to ensure real-time power balance and power quality in the power grid, effective data support and decision support are needed for generation-side regulation based on system operational status. Power-side adjustment information, including key information on the regulation characteristics of various generator units, output recommendations, and response timing, can be generated based on multi-dimensional data such as load forecasting and frequency stability assessment. This information provides a reference for power-side managers in their operations within the energy management system. Therefore, power-side managers can adjust generation plans and power output based on the integrated analysis results.

[0057] For example, after identifying operational trends such as "load will increase by 5%" and "insufficient frequency margin," key information can be obtained, such as the rapid adjustment capability of hydropower station A, the suitability of gas-fired power plant B for medium-term load tracking, and the ability of thermal power plant C to participate in frequency stability support. Through this processing, a complete decision support data package for optimized scheduling on the power supply side is formed, providing information for flexible adjustment of generator output and frequency control.

[0058] To effectively utilize the rapid and flexible characteristics of energy storage systems to smooth grid fluctuations and alleviate transmission congestion, data support and decision support for their charging and discharging operations are needed based on the system's operating status. In this scenario, multi-dimensional data such as line power flow, frequency information, load forecasting, and renewable energy output can be combined to generate control strategy recommendations, including energy storage site recommendations, timing of actions, power range information, and duration information, assisting grid-side personnel in making judgments. For example, upon identifying operating conditions such as "power flow at section XX is approaching its limit" or "system frequency remains persistently low," corresponding strategy prompts can be output, suggesting charging the M energy storage station on the sending end to alleviate congestion, or utilizing energy storage resources for rapid frequency regulation during frequency drops. This generates decision support information for optimized operation of the energy storage system, providing a reliable basis for flexible grid-side control. Thus, grid-side managers can optimize the charging and discharging of the energy storage system based on the fusion analysis results.

[0059] Furthermore, based on the fusion analysis results, the historical operation and maintenance data of a specified node can be processed to remove impurities and obtain the operation and maintenance data quality information of the specified node, so that load-side managers can formulate demand response strategies based on the fusion analysis results.

[0060] The impurity removal process can be expressed as follows: , in, For operational data quality information, Let be the coordinates of a parameter point at any node in the distribution network. To efficiently maintain the amount of data, To improve the efficiency of static management of operation and maintenance data, To ensure efficient operation and maintenance, the number of commands issued should be [number]. For the first The static management efficiency of issuing efficient operation and maintenance commands. It is a collection of original, efficient operation and maintenance data.

[0061] For example, assuming five historical operation and maintenance data records are collected for a certain node, the static management efficiency is... Represented as: , Set the static management efficiency of a certain high-efficiency operation and maintenance command. It is 92%; based on and Substituting the values ​​into the calculation yields the following results: , As can be seen from the above, The threshold is 1.4%. A data quality threshold (e.g., 0) can be preset. Since the data quality information is greater than the data quality threshold, i.e., 1.4%>0, it indicates that the overall data quality of the corresponding node is higher than the baseline, the data set is relatively pure, and it can be used to train the aforementioned Transformer model or support decision-making.

[0062] Therefore, based on the results of the integrated analysis of users' electricity consumption patterns, load-side managers can formulate demand response strategies and remove impurities from relevant operation and maintenance data, which can provide reliable data support and decision support for tapping the load-side adjustment potential and achieving "peak shaving and valley filling".

[0063] Furthermore, by analyzing user load characteristics, interruptibility potential, and response willingness, key auxiliary information can be output, including user classification, adjustment potential assessment, incentive recommendations, and recommended execution periods. This supports load-side managers in guiding grid interaction through advanced measurement systems or virtual power plant platforms. For example, after identifying the risk of load surges caused by extreme heat, adjustment potential assessments for different user groups can be output, providing data support for interactive strategies such as air conditioning control in large shopping malls or guiding electric vehicle charging in residential areas. This can form demand response decision support information covering user types, response methods, and time windows, laying the foundation for flexible load-side adjustment and precise interaction.

[0064] Figure 4 A schematic diagram of the structure of a power grid data fusion system provided in an embodiment of this disclosure is shown. See also Figure 4 The power grid data fusion system provided in this disclosure embodiment may include: The data acquisition unit 401 is used to acquire power grid data collected by SCADA, WAMS and RPMS to generate a multi-source dataset; The fusion analysis unit 402 is used to obtain fusion analysis results using multi-source datasets. The fusion analysis results include comprehensive operating indicators and abnormal information. The comprehensive operating indicators are obtained through principal component analysis, and the abnormal information is obtained through information identification and extraction. Result providing unit 403 is used to provide the fusion analysis results to the application terminal.

[0065] Furthermore, the data acquisition unit 401 can be specifically used to: align, label, and structure the second-level steady-state data generated by SCADA acquiring power grid operation data collected by sensors at each node of the power grid, the millisecond-level synchronization phasor data generated by WAMS acquiring power grid operation data collected by sensors at each node of the power grid, and the fault event data generated by RPMS acquiring power grid operation data collected by sensors at each node of the power grid, based on a preset standardized timestamp, to obtain a multi-source dataset.

[0066] Furthermore, the power grid data fusion system provided in this embodiment may further include: a data preprocessing unit 404, used to preprocess multi-source datasets, the preprocessing including one or more of the following: removing duplicate data, filling in missing data, normalization processing, and standardization processing.

[0067] The power grid data fusion system of this disclosure can be implemented by software, hardware, or a combination of both. For example, the power grid data fusion system of this disclosure can be implemented as software running in the electronic device 600 described below.

[0068] Figure 5 A software architecture example diagram of the power grid data fusion system provided in this disclosure embodiment is shown. See also... Figure 5 The power grid data fusion system of this disclosure can adopt a three-layer collaborative architecture of data acquisition layer, data management layer, and application layer to provide systematic optimization for the operation and management of power grid enterprises. The data acquisition unit 401 is located in or can be implemented as the data acquisition layer, the data preprocessing unit 404 and the fusion analysis unit 402 can be located in the data management layer, and the result providing unit 403 can be located in the application layer. The data acquisition layer integrates data from multiple systems such as SCADA, WAMS, and RPMS to form a multi-source dataset, ensuring the comprehensiveness and accuracy of data acquisition and breaking down information silos at the source. The data management layer can perform data preprocessing, feature extraction, and intelligent analysis, completing deep fusion and proactive mining of multi-source datasets, significantly improving the value density and usability of the data. At the application layer, the fusion analysis results can be provided to application terminals to provide data support and decision support for user terminals on the power supply side, grid side, and load side. Through the power grid data fusion of this disclosure, not only can the execution efficiency of power generation planning, energy storage scheduling, and demand response be improved, but cross-departmental resource optimization and process collaboration can also be achieved.

[0069] See Figure 5The data acquisition layer collects real-time power grid enterprise operation data from various nodes through sensors, SCADA, WAMS, and RPMS to obtain multi-source datasets. The data management layer performs data preprocessing, feature extraction, and key information identification on the multi-source datasets to obtain fusion analysis results containing comprehensive operation indicators and abnormal information, realizing fusion analysis of multi-source data and intelligent perception of enterprise operation status, and obtaining fusion analysis results. Based on the fusion analysis results, the application layer provides data support and decision support for power supply, grid, and load side management personnel, enabling enterprise operation management personnel to formulate control strategies in combination with the actual operation status of the power grid and business needs.

[0070] Figure 6 A schematic structural diagram of an electronic device provided according to an embodiment of this disclosure is shown. See also... Figure 6 The electronic device 600 provided in this embodiment may include one or more processors 601 and a memory 602. The memory 602 stores a computer program, which, when run by the processor 601, causes the processor 601 to execute the above-described power grid data fusion method.

[0071] The processor 601 may be a central processing unit (CPU), a graphics processing unit (GPU), or other form of processing unit with data processing capabilities and / or instruction execution capabilities, and may control other components in the computing unit to perform desired functions.

[0072] The memory 602 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and a processor may execute the program instructions to implement the methods of the various embodiments of this disclosure described above and / or other desired functions.

[0073] Depending on the specific application, the electronic device 600 may also include any other suitable components.

[0074] In addition to the methods and devices described above, embodiments of this disclosure may also be computer program products, including computer program instructions that, when executed by a processor, cause the processor to perform the steps of the power grid data fusion method described in the embodiments of this disclosure.

[0075] The computer program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of this disclosure. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on a user's computing device, partially on a user's computing device, as a standalone software package, partially on a user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0076] Furthermore, embodiments of this disclosure may also be computer-readable storage media storing computer program instructions thereon, which, when executed by a processor, cause the processor to perform the steps of the power grid data fusion method described in embodiments of this disclosure.

[0077] The computer-readable storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0078] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this disclosure to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations therein.

Claims

1. A power grid data fusion method, characterized in that, The method includes: Acquire power grid data collected by SCADA, WAMS, and RPMS to generate a multi-source dataset; The multi-source dataset is used to obtain fusion analysis results, which include comprehensive operating indicators and anomaly information. The comprehensive operating indicators are obtained through principal component analysis, and the anomaly information is obtained through anomaly identification. The fusion analysis results are provided to the application terminal.

2. The method according to claim 1, characterized in that, The process of acquiring power grid data collected by the SCADA system, WAMS system, and RPMS system to generate a multi-source dataset includes: Sensors deployed at each node of the power grid collect power grid operation data for the corresponding node. SCADA acquires power grid operation data collected by sensors at various nodes of the power grid to generate second-level steady-state data, which includes both digital and analog signals. WAMS acquires power grid operation data collected by sensors at each node of the power grid to generate millisecond-level synchronous phasor data. The millisecond-level synchronous phasor data includes the amplitude and phase angle of the voltage fundamental phasor and the amplitude and phase angle of the current fundamental phasor. RPMS acquires power grid operation data collected by sensors at each node of the power grid to generate fault event data. The fault event data includes protection trip commands, switch quantity change sequences, and instantaneous waveforms of current and voltage before and after the fault. The second-level steady-state data, the millisecond-level synchronization phasor data, and the fault event data are aligned, labeled, and structured and stored based on a preset standardized timestamp to obtain the multi-source dataset.

3. The method according to claim 1, characterized in that, The method further includes: Before obtaining the fusion analysis results using the multi-source dataset, the multi-source dataset is preprocessed. The preprocessing includes one or more of the following: removing duplicate data, filling in missing data, normalization, and standardization.

4. The method according to claim 1, characterized in that, The comprehensive operating indicators are obtained in the following way: The multi-source dataset is standardized to eliminate dimensions and obtain the original data matrix; The covariance matrix is ​​calculated using the original data matrix. The number of principal components is determined by solving the eigenvalues ​​and eigenvectors of the covariance matrix, the contribution rate of each principal component is calculated, the cumulative contribution rate of multiple principal components is calculated, and the principal component score matrix is ​​constructed. The principal component score matrix is ​​the comprehensive operating index.

5. The method according to claim 1 or 4, characterized in that, The anomaly information and the comprehensive operating indicators are obtained through a pre-trained Transformer model, which is used to perform the principal component analysis and the anomaly identification in parallel.

6. The method according to claim 1, characterized in that, After providing the fusion analysis results to the application terminal, the method further includes one or more of the following: Frequency stability assessment data is obtained based on the fusion analysis results; power supply adjustment information is generated based on the frequency stability assessment data and the load forecast data obtained from the power supply side application terminal, the power supply adjustment information includes key information on the adjustment characteristics of various generator sets, output suggestion information and response timing; Frequency information is obtained based on the fusion analysis results; Based on the frequency information and the line power flow information, load forecast information and new energy output information obtained by the second application terminal, control strategy suggestion information is generated. The control strategy suggestion information includes energy storage site recommendation information, action timing information, power range information and duration information. Based on the fusion analysis results, the historical operation and maintenance data of the specified node is processed to remove impurities in order to obtain the operation and maintenance data quality information of the specified node.

7. A power grid data fusion system, characterized in that, The power grid data fusion system includes: The data acquisition unit is used to acquire power grid data collected by SCADA, WAMS and RPMS to generate multi-source datasets; The fusion analysis unit is used to obtain fusion analysis results using the multi-source dataset. The fusion analysis results include comprehensive operating indicators and abnormal information. The comprehensive operating indicators are obtained through principal component analysis, and the abnormal information is obtained through information identification and extraction. The result providing unit is used to provide the fusion analysis results to the application terminal.

8. The apparatus according to claim 7, characterized in that, The data acquisition unit is specifically used to: align, label, and structure the second-level steady-state data generated by the SCADA acquiring power grid operation data collected by sensors at each node of the power grid, the millisecond-level synchronization phasor data generated by the WAMS acquiring power grid operation data collected by sensors at each node of the power grid, and the fault event data generated by the RPMS acquiring power grid operation data collected by sensors at each node of the power grid, based on a preset standardized timestamp, to obtain the multi-source dataset.

9. The apparatus according to claim 7, characterized in that, The power grid data fusion system further includes a data preprocessing unit, used to preprocess the multi-source dataset, wherein the preprocessing includes one or more of the following: removing duplicate data, filling in missing data, normalization processing, and standardization processing.

10. An electronic device comprising a processor and a memory, the memory storing a computer program that, when executed by the processor, causes the processor to perform the method as described in any one of claims 1-6.

Images