Weak support large-scale new energy delivery system multi-source state trend integrated data fusion method

Abstract

The invention discloses a multi-source state trend integrated data fusion method for a weak support large-scale new energy delivery system. According to the invention, a data source identification and hierarchical access mechanism, a multi-source synchronization and calibration method based on a unified time reference, a topology-oriented space mapping method, a trend feature extraction method based on a sliding window and multi-scale analysis, and a state-trend integrated weak support risk characterization index system are constructed; unified modeling and online sensing of key quantities such as voltage supporting capacity, power flow margin, standby level, renewable output fluctuation and the like are realized. Compared with an existing monitoring mode depending on a single system, a single time scale or a single index, the method has the advantages that on the basis of ensuring data quality and time-space consistency, a unified data base and trend sensing capability can be provided for safety evaluation, cooperative regulation and market linkage control of the weak-support large-scale new energy delivery system; and the method has relatively high engineering applicability and popularization value.

Description

Technical Field

[0001] This invention belongs to the field of power system operation and control technology, specifically relating to a multi-source status and trend integrated data integration method for large-scale renewable energy transmission systems with weak support, used to support real-time decision-making for transmission channel security assessment, active and reactive power coordinated control, and linkage with the spot electricity market. Background Technology

[0002] In scenarios involving high-proportion renewable energy integration and long-distance transmission, transmission channels exhibit typical weak support characteristics due to factors such as insufficient short-circuit capacity of the receiving-end power grid, weak reactive power support capabilities, and rigid power flow distribution. In such systems, the multi-source status and trend information of wind power, photovoltaics, energy storage, grid-side reactive power compensation devices, and the transmission channels themselves are highly coupled and span different voltage levels, time scales, and control entities.

[0003] In existing technologies, on the one hand, various operation monitoring and control systems (such as power plant SCADA, dispatch EMS, WAMS / PMU, protection and safety automatic devices, frequency regulation and peak shaving control systems, market trading and settlement systems, etc.) are often built independently, with inconsistent data interface standards and significant differences in sampling periods, making it difficult to form a unified state and trend view; on the other hand, most monitoring methods only focus on a certain type of state quantity (such as voltage, power angle, cross-sectional power flow or price series), lacking systematic trend feature construction and cross-source fusion methods oriented towards weak support mechanisms.

[0004] In scenarios involving large-scale renewable energy transmission with weak support, without high-quality access, unified alignment, and deep integration of multi-source status and trend data, it becomes difficult to accurately identify key issues such as voltage instability risks, power flow overrun risks, insufficient reserves, and significant fluctuations in renewable energy. Furthermore, it hinders the provision of quantitative data for multi-site collaborative control and market-driven scheduling. Therefore, a method for accessing and integrating multi-source status and trend data that balances heterogeneity, spatiotemporal consistency, and the constraints of weak support mechanisms is urgently needed. Summary of the Invention

[0005] To address the problems existing in the prior art, the purpose of this invention is to provide a multi-source state trend integrated data fusion method for a weakly supported large-scale renewable energy transmission system, which combines multi-site joint market response capability, power grid risk perception capability, and control feedback execution capability, thereby achieving integrated control of market, dispatch, and economic mechanisms.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A method for integrating multi-source state and trend data in a weakly supported large-scale renewable energy transmission system includes the following steps: Step S1: Identification of multiple data sources and planning of hierarchical access architecture; Step S2: Multi-source state and time series data collection, preprocessing and quality labeling; Step S3: Time synchronization based on unified clock and spatial mapping based on topology; Step S4: Multi-scale trend feature construction and weak support sensitive feature extraction; Step S5: State-trend data fusion and weak support risk indicator system construction; Step S6: Trend consistency evaluation and anomaly detection in weak support scenario; Step S7: Unified data service and interface publishing for regulation and market application; Among them, each step is executed in sequence or partially in parallel through modular design in engineering application.

[0007] Step S1 is used to systematically sort out the data source types, attributes and importance in the overall architecture of the weakly supported large-scale new energy sending system, and form a hierarchical access architecture plan; (1) Data source classification and modeling The data sources are divided into three categories: power generation side source data, power grid side source data, and market and weather source data. The power generation side source data includes SCADA measurement and control instructions of wind power, photovoltaic, energy storage, and conventional generating units. The power grid side source data includes substation, voltage support device, transmission section power flow, PMU and WAMS measurement protection, and automation action record. The market and weather source data includes spot and medium and long-term electricity price sequence, clearing quantity, dispatching plan, power prediction curve and multi-dimensional weather elements; For each type of data source, a unified source-level description model is established, and its sampling period Δt, spatial attribution, physical dimension and quality label set Q are defined; (2) Hierarchical access architecture planning The access layer is divided into edge collection layer, main trunk aggregation layer and center integration layer. The edge collection layer is responsible for establishing a secure channel with the field device and the boundary system, the main trunk aggregation layer realizes data aggregation and buffering across voltage levels and dispatching levels, and the center integration layer is responsible for unified modeling and service. By specifying the access level, interface protocol and update strategy for each data source, it is ensured that the key state and trend data in the weak support sending scenario have sufficient timeliness and reliability.

[0008] Step S2 is as follows: Step S2 is aimed at the planned data sources, and implements standardized collection and preprocessing, and provides quality guarantee for subsequent trend construction and risk assessment; (1) Collection and buffering mechanism Establish a timestamp-based streaming acquisition channel, sample high-frequency PMU data at the millisecond or tens of millisecond level, sample SCADA and market data at the second to minute level, and achieve reliable data transmission and short-term caching through circular buffer or message queue mechanism; (2) Data cleaning and initial screening of anomalies For the collected time series S i (t) Perform outlier detection, missing value identification, and duplicate value removal operations; (3) Quality markers and pedigree records Add a quality label q to each raw or preprocessed data record i (t)∈Q, and record the data collection time, processing flow identifier and source system identifier to form a traceable data spectrum; the quality label will be used for weighting or elimination strategies in subsequent trend feature construction and weak support indicator calculation, so as to avoid low-quality data from misleading risk assessment.

[0009] Preferably, outlier detection employs the 3σ criterion, box plot method, or threshold determination based on historical probability distribution; missing values ​​are filled in using interpolation between adjacent time points or regression estimation based on adjacent correlation quantities. Step S3 is as follows: In scenarios with weak support for data transmission, the temporal and spatial consistency of multi-source data is crucial for trend analysis and mechanism characterization. Step S3 is to achieve multi-source data alignment under a unified clock and unified topology framework. (1) Unified time base and time scale correction Introducing a unified time base T0, and applying timestamps t from various data sources i To perform correction, for data sources with clock drift or communication delay, the systematic deviation δt is estimated and compensated by calculating the correlation coefficient or the time difference of synchronization events with the reference source. i Thus, the corrected time t is obtained. i =t i δt i ; (2) Multi-scale time alignment and resampling For time series with different sampling periods, they will be uniformly mapped to a set of nested time scales {ΔT1,ΔT2,…}. The original sequence will be mapped to different scales that express the state and trend through resampling or aggregation. In the assessment of weak support, the details of rapid changes in voltage and frequency will be maintained at the second level, while the power flow, reserve and market clearing volume will show trends at the minute level. (3) Topology-based spatial mapping and equivalence A network topology model of the power transmission system is constructed, including power generation stations, step-up substations, transmission lines, key receiving-end sections, and receiving-end main grid nodes. Measurement point codes in each data source are mapped to topology node or branch numbers, and data from different systems located at the same electrical node are spatially aligned. For some quantities that cannot be directly mapped, they are mapped to a unified topology region or equivalent node through power flow equivalence or topology folding methods, laying the foundation for the subsequent construction of weak support indicators at multiple levels of nodes-sections-regions.

[0010] Step S4 is as follows: Step S4, based on the spatiotemporal alignment of state data, constructs multi-scale trend features for weak support mechanisms and mines the feature quantities most sensitive to voltage support and transmission security. (1) Sliding window trend construction At a given time scale ΔT k Below, regarding the key variable x i (t) Construct a sliding window W k (t)=[t L k ,t], calculate the mean μ ik (t), slope g ik (t), volatility σ ik (t) Trend quantity; slope is expressed as: g ik (t)= [x i (t) x i (t L k )] / L k , among which, L k Indicates the duration of the sliding window; Volatility is measured using the standard deviation within a window or the absolute deviation. (2) Multiscale decomposition and frequency band characteristics For high-frequency measurement data of PMU voltage and frequency, discrete wavelet transform or empirical mode decomposition is used to decompose the signal into components of different frequency bands, and low-frequency trend and high-frequency disturbance characteristics are extracted. The low-frequency components are used to characterize the slowly changing power flow and voltage evolution, while the high-frequency components are used to characterize transient disturbances and oscillations, thereby reflecting the multi-timescale dynamic characteristics of the weakly supported system under a unified framework. (3) Weak support sensitive feature extraction Based on sensitivity analysis theory, the sensitivity of node voltage V to injected power ΔP and ΔQ is analyzed. V / P, V / Q. Sensitivity of power flow at key cross-sections to fluctuations in renewable energy output and load. P line / P nen , P line / P loa d performs calculations and constructs a sensitive trend quantity in the trend dimension. S vp (t)= ( V / P) · g p (t), S v _q(t)= ( V / Q) · g_q(t), Where g p (t) and g_q(t) are the trend slopes of the corresponding power injection; S vp (t), S v _q(t) is a sensitive trend indicator for assessing the vulnerability of node voltage; the above sensitive trend characteristics can intuitively reflect the vulnerability of node voltage under the current fluctuation trend.

[0011] Key variable x i (t) includes key node voltage V, active power flow P, reactive power output Q, and renewable energy output P. nen And reserve margin R.

[0012] Step S5 is as follows: Step S5: Based on multi-source and multi-scale trend features, construct a unified state-trend fusion dataset and a weak support risk characterization index to provide a quantitative basis for the assessment and control of outbound security. (1) Construction of unified feature vector of state-trend For each key analysis object, construct an feature vector containing instantaneous state variables and trend variables at a unified time grid point t: z(t) = [V(t), P(t), Q(t), R(t), μ(t), g(t), σ(t), S vp (t), S v _q(t), …] Where V(t), P(t), Q(t), and R(t) represent instantaneous voltage, active power flow, reactive power support, and reserve margin, respectively, and μ(t), g(t), and σ(t) represent trend statistics under different window scales; (2) Design of comprehensive indicators for weak support Based on the eigenvector z(t), a weak support comprehensive index I_w(t) is constructed, using a weighted linear or nonlinear combination: I_w(t) = w1·I_V(t) + w2·I_P(t) + w3·I_Q(t) + w4·I_R(t) + w5·I_trend(t) Where I_V(t) is based on voltage deviation and voltage margin, I_P(t) is based on the cross-sectional power flow ratio, I_Q(t) is based on the utilization level of non-functional capacity, I_R(t) is based on the consumption level of reserve margin, and I_trend(t) is constructed based on trend slope and volatility; weight w i Determined through training with historical data or expert experience; (3) Indicator stratification and alarm threshold I_w(t) is mapped to several risk levels, namely normal, mild weak support, severe weak support, and critical instability, with different color labels and alarm strategies. Differentiated thresholds are set for key nodes or sections to form a multi-level and multi-dimensional weak support risk map, providing a basis for subsequent regulation and market price linkage.

[0013] Step S6 is as follows: Since the weak support system is sensitive to data quality, a trend consistency assessment and anomaly detection mechanism is introduced in step S6 to ensure the reliability of the state-trend fusion data; (1) Physical consistency verification Based on the power flow balance constraint and energy conservation relationship, the consistency of multi-source state variables is checked to examine the balance relationship between station output, plant power, external power transmission and network loss; whether the changes in node voltage and reactive power output conform to the static characteristic curve; if the observed trend violates the physical constraints, the corresponding data segment is marked as abnormal and its weight in the calculation of weak support index is reduced. (2) Correlation analysis between trends By calculating the correlation coefficients or mutual information between the trends of different variables, namely the trend of new energy output and the trend of cross-sectional power flow, the trend of voltage and the trend of reactive power output, and the trend of electricity price and the trend of power flow, abnormal combinations inconsistent with the weak support mechanism can be identified. (3) Anomaly detection and automatic handling Anomaly detection is performed on the state-trend feature vector z(t) using methods based on statistical thresholds, cluster distance, or simple machine learning models. Data segments identified as anomalous are automatically processed, i.e., marked, removed, or reconstructed by interpolating data from adjacent time periods and neighboring nodes, while retaining the original records for post-analysis and auditing.

[0014] Step S7 is as follows: Based on the above access and integration steps, step S7 constructs a unified data service layer to provide standardized interfaces for external security assessment, collaborative regulation and market linkage. (1) Unified data view and topic view A unified data view is constructed around the hierarchical structure of nodes, sections, channels, regions, and markets, providing a comprehensive display of the status, trends, and weak support indicators for key objects; at the same time, a security assessment view, a new energy fluctuation view, a channel margin view, and a price-trend linkage view are constructed to meet the needs of different applications; (2) Standardized interface and subscription mechanism The design incorporates standardized query and subscription interfaces, allowing external applications to subscribe to the required data based on time intervals, spatial ranges, and metric types. The data service layer supports both real-time push and historical playback modes, while retaining access logs and call records to meet security auditing and performance optimization requirements. (3) Coupling method between control and market decision-making modules Without altering the main control logic of the existing scheduling and control system, weak support indicators and trend information are embedded as constraints or penalties into the output-pricing decision model for power transmission control, reactive power optimization, and participation in the spot and ancillary service markets. This achieves a systematic linkage between enhanced perception on the data side, explicit constraints on the model side, and closed-loop execution on the control side.

[0015] This invention addresses the problems encountered during the transmission of high-proportion renewable energy bases, such as those in desert areas, where there are inconsistent standards for collecting multi-source heterogeneous data (SCADA, PMU / WAMS, automatic protection and safety devices, EMS dispatching, meteorological data, and market transactions), time scale inconsistencies, spatial topology misalignment, and separation of state and trend quantities. It proposes a technical approach for unified access, alignment, fusion, and indicator construction of multi-source state and trend data in scenarios with weak support for power transmission. Compared with existing technologies, this invention has the following advantages: Multi-source fusion for weak support mechanisms: By systematically accessing state and trend data from the generation, grid, and market sides and modeling them under a unified spatiotemporal and topological framework, the limitations of traditional single-system, single-time-scale monitoring are overcome; Integrated state-trend weak support indicators: By introducing multi-scale trend characteristics and sensitivity analysis, the instantaneous state and trend evolution process are uniformly expressed, constructing a comprehensive indicator that can directly characterize voltage support vulnerability and channel safety margin; Controllable and traceable data quality: Through quality labeling, spectral recording, trend consistency assessment, and anomaly detection mechanisms, the interpretability and reliability of the data on which weak support risk assessment depends are ensured; On the basis of ensuring data quality and spatiotemporal consistency, a unified data foundation and trend perception capability are provided for the safety assessment, coordinated regulation, and market linkage control of large-scale renewable energy transmission systems with weak support, facilitating engineering implementation and expansion: Without relying on specific big data platforms or simulation tools, this invention adopts a layered access, standardized modeling, and interface release approach, which can be smoothly superimposed on the existing scheduling and control systems and is also easy to integrate with advanced applications such as multi-site collaborative control and market bidding optimization. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0018] This invention provides a multi-source state trend integrated data fusion method for a weakly supported large-scale renewable energy transmission system, comprising the following main steps: Step S1: Identification of multiple data sources and planning of hierarchical access architecture; Step S2: Acquisition, preprocessing, and quality labeling of multi-source state and time series data; Step S3: Time synchronization based on a unified clock and spatial mapping based on topology; Step S4: Construction of multi-scale trend features and extraction of weak support sensitive features; Step S5: Integration of Status and Trend Data and Construction of a Weak Support Risk Indicator System; Step S6: Trend consistency assessment and anomaly detection in weak support scenarios; Step S7: Release of unified data services and interfaces for regulation and market applications.

[0019] The steps can be executed sequentially, or in engineering applications, they can be partially parallelized through modular design. The following provides a detailed academic and engineering description of each step.

[0020] Step S1 is used to systematically sort out the data source types, attributes and importance within the overall architecture of a large-scale new energy transmission system with weak support, and form a hierarchical access architecture plan.

[0021] (1) Data source classification and modeling Data sources are categorized into three main types: generation-side data, grid-side data, and market and meteorological data. Generation-side data includes SCADA measurements and control commands from wind power, photovoltaic, energy storage, and conventional generating units; grid-side data includes data from substations, voltage support devices, power flow at transmission sections, PMU / WAMS measurements and protection, and automation operation records; market and meteorological data includes spot / medium- to long-term electricity price series, clearing volumes, dispatch plans, power forecast curves, and multi-dimensional meteorological elements.

[0022] For each type of data source, a unified source-level description model is established, defining its sampling period Δt, spatial affiliation (station, line, section, region), physical dimensions (voltage, current, active power, reactive power, temperature, wind speed, etc.) and quality label set Q (such as real-time, post-correction, missing interpolation, manual correction, etc.).

[0023] (2) Layered access architecture planning The access layer is divided into three parts: an edge acquisition layer, a backbone aggregation layer, and a central integration layer. The edge acquisition layer is responsible for establishing secure channels with field devices and boundary systems; the backbone aggregation layer enables data aggregation and buffering across voltage levels and scheduling levels; and the central integration layer is responsible for unified modeling and services. By specifying the access layer, interface protocol, and update strategy for each data source, sufficient timeliness and reliability of critical status and trend data are ensured in scenarios with weak support for data transmission.

[0024] Step S2 is as follows: Step S2 involves standardized data collection and preprocessing based on the planned data sources, providing quality assurance for subsequent trend construction and risk assessment.

[0025] (1) Acquisition and buffering mechanism Establish a timestamp-based streaming acquisition channel, sample high-frequency PMU data at the millisecond or tens of millisecond level, and sample SCADA and market data at the second to minute level. Reliable data transmission and short-term caching are achieved through circular buffers or message queue mechanisms.

[0026] (2) Data cleaning and initial screening of anomalies For the collected time series S i(t) Perform preprocessing operations such as outlier detection, missing value identification, and duplicate value elimination. Outlier detection can be performed using the 3σ criterion, box plot method, or threshold determination based on historical probability distribution; missing values ​​can be reasonably filled by interpolation based on adjacent time points or regression estimation based on adjacent related quantities (such as power flow and voltage of adjacent lines).

[0027] (3) Quality markers and pedigree records Add a quality label q to each raw or preprocessed data record i (t)∈Q, and record the data collection time, processing flow identifier, and source system identifier to form a traceable data spectrum. Quality labels will be used in subsequent trend feature construction and weak support indicator calculation weighting or elimination strategies to avoid low-quality data from misleading risk assessment.

[0028] Step S3 is as follows: In scenarios with weak data transmission support, the temporal and spatial consistency of multi-source data is crucial for trend analysis and mechanism characterization. Step S3 involves aligning multi-source data under a unified clock and topology framework.

[0029] (1) Unified time base and time scale correction Introduce a unified time base T0 (such as GPS or BeiDou time synchronization) and timestamp t from each data source. i Correction is performed. For data sources with clock drift or communication delay, the systematic deviation δt is estimated and compensated by calculating the correlation coefficient with the reference source (such as a PMU) or the time difference of synchronization events (such as switching actions). i Thus, the corrected time t is obtained. i =t i δt i .

[0030] (2) Multi-scale time alignment and resampling For time series with different sampling periods, they will be uniformly mapped to a set of nested time scales {ΔT1, ΔT2, ...}. For example, using ΔT1=1s, ΔT2=1min, ΔT3=15min, etc., as multi-scale time grids, the original sequence can be mapped to different scales expressing states and trends through resampling or aggregation. In weak support assessment, it can ensure that rapidly changing quantities such as voltage and frequency retain details at the second level, while power flow, reserve, and market clearing volumes show trends at the minute and 15-minute levels.

[0031] (3) Topology-based spatial mapping and equivalence A network topology model of the power transmission system is constructed, including power plants, step-up substations, transmission lines, key receiving-end sections, and receiving-end main grid nodes. Measurement point codes in each data source are mapped to topology node or branch numbers, and data from different systems located at the same electrical node are spatially aligned. For quantities that cannot be directly mapped (such as regional aggregated power), power flow equivalence or topology folding methods are used to map them to a unified topology region or equivalent nodes, laying the foundation for subsequent construction of weak support indicators at a multi-level node-section-region structure.

[0032] Step S4 is as follows: Step S4: Based on the spatiotemporal alignment of state data, construct multi-scale trend features for weak support mechanisms and mine the feature quantities most sensitive to voltage support and transmission security.

[0033] (1) Sliding window trend construction At a given time scale ΔT k Below, regarding the key variable x i (t) Construct a sliding window W k (t)=[t L k ,t], calculate the mean μ ik (t), slope g ik (t), volatility σ ik (t) is an isotropic quantity. For example, the slope can be expressed as: g ik (t)= [x i (t) x i (t L k )] / L k , among which, L k Indicates the duration of the sliding window; Volatility can be measured using the standard deviation within a window or the absolute deviation.

[0034] (2) Multiscale decomposition and frequency band characteristics For high-frequency measurement data (such as PMU voltage and frequency), discrete wavelet transform or empirical mode decomposition can be used to decompose the signal into components of different frequency bands, extracting low-frequency trends and high-frequency disturbance characteristics. Low-frequency components are used to characterize slowly varying power flow and voltage evolution, while high-frequency components are used to characterize transient disturbances and oscillations, thus reflecting the multi-timescale dynamic characteristics of the weakly supported system within a unified framework.

[0035] (3) Weak support sensitive feature extraction Based on sensitivity analysis theory, the sensitivity of node voltage V to injected power ΔP and ΔQ is analyzed. V / P, V / Q. Sensitivity of power flow at key cross-sections to fluctuations in renewable energy output and load. P line / P nen , P line / P loa d, etc., are calculated, and sensitive trend quantities are constructed in the trend dimension, for example: S vp (t)= ( V / P) · g p (t), S v _q(t)= ( V / Q) · g_q(t), Where g p (t) and g_q(t) are the trend slopes of the corresponding power injection; S vp (t), S v _q(t) is a sensitive trend indicator for assessing the vulnerability of node voltage. The above sensitive trend characteristics can intuitively reflect the vulnerability of node voltage under the current fluctuation trend.

[0036] Step S5 is as follows: Step S5: Based on multi-source and multi-scale trend features, construct a unified state-trend fusion dataset and weak support risk characterization indicators to provide quantitative basis for the assessment and control of outbound security.

[0037] (1) Construction of unified feature vector of state-trend For each key analysis object (such as a node, cross section, region, or transmission channel), construct an feature vector containing instantaneous state and trend quantities at a unified time grid point t: z(t) = [V(t), P(t), Q(t), R(t), μ(t), g(t), σ(t), S vp (t), S v _q(t), …] Where V(t), P(t), Q(t), and R(t) represent instantaneous voltage, active power flow, reactive power support, and reserve margin, respectively, and μ(t), g(t), σ(t), etc. represent trend statistics under different window scales.

[0038] (2) Design of comprehensive indicators for weak support Based on the eigenvector z(t), a weak support comprehensive index I_w(t) is constructed, for example, by using a weighted linear or nonlinear combination: I_w(t) = w1·I_V(t) + w2·I_P(t) + w3·I_Q(t) + w4·I_R(t) + w5·I_trend(t) Where I_V(t) can be based on voltage deviation and voltage margin, I_P(t) on the cross-sectional power flow ratio, I_Q(t) on the degree of non-functional capacity utilization, I_R(t) on the degree of reserve margin consumption, and I_trend(t) is constructed based on trend slope and volatility. Weight w i It can be determined through training with historical data or expert experience.

[0039] (3) Indicator stratification and alarm threshold I_w(t) is mapped to several risk levels (such as normal, mild weak support, severe weak support, and critical instability), corresponding to different color codes and alarm strategies. Differentiated thresholds can be set for key nodes or sections to form a multi-level, multi-dimensional weak support risk map, providing a basis for subsequent regulation and market price linkage.

[0040] Step S6 is as follows: Since weak support systems are sensitive to data quality, this invention introduces a trend consistency assessment and anomaly detection mechanism in step S6 to ensure the reliability of state-trend fusion data.

[0041] (1) Physical consistency verification Based on power flow balance constraints and energy conservation, consistency checks are performed on multi-source state variables. For example, the balance between station output, plant power consumption, transmitted power, and network losses is checked; whether node voltage and reactive power output changes conform to the static characteristic curve is also checked. If the observed trend clearly violates physical constraints, the corresponding data segment is marked as an anomaly and its weight in the calculation of weak support index is reduced.

[0042] (2) Correlation analysis between trends By calculating the correlation coefficients or mutual information between different variable trends (such as the trend of renewable energy output and the trend of cross-sectional power flow, the trend of voltage and the trend of reactive power output, and the trend of electricity price and the trend of power flow), abnormal combinations inconsistent with the weak support mechanism can be identified. For example, if the transmitted power is rising for a long period of time but the voltage trend continues to rise, there may be data anomalies or measurement errors if there is a lack of corresponding changes in reactive power support.

[0043] (3) Anomaly detection and automatic handling Anomaly detection is performed on the state-trend feature vector z(t) using methods based on statistical thresholds, cluster distance, or simple machine learning models (such as isolated forests). Data segments identified as anomalous are automatically processed, such as marked, removed, or reconstructed using data from neighboring time periods and adjacent nodes, while the original records are retained for post-analysis and auditing.

[0044] Step S7 is as follows: Based on the above access and integration steps, the present invention constructs a unified data service layer in step S7 to provide standardized interfaces for external security assessment, collaborative regulation and market linkage.

[0045] (1) Unified data view and topic view A unified data view is constructed around the hierarchical structure of nodes, sections, channels, regions, and markets, providing a comprehensive display of the status, trends, and weak support indicators for key objects. At the same time, thematic views such as security assessment view, new energy fluctuation view, channel margin view, and price-trend linkage view are constructed to meet the needs of different applications.

[0046] (2) Standardized interface and subscription mechanism The system features standardized query and subscription interfaces, allowing external applications (such as active power collaborative control systems, multi-site joint bidding optimization platforms, and spot market analysis systems) to subscribe to the required data by time interval, spatial range, and indicator type. The data service layer supports both real-time push and historical playback modes, while retaining access logs and call records to meet security auditing and performance optimization requirements.

[0047] (3) Coupling method with control and market decision-making modules Without altering the main control logic of the existing scheduling and control system, the weak support indicators and trend information provided by this invention are embedded as constraints or penalties into the output-pricing decision model for power transmission control, reactive power optimization, and participation in the spot and ancillary service markets, thereby achieving systematic linkage between enhanced perception on the data side, explicit constraints on the model side, and closed-loop execution on the control side.

[0048] The present invention also discloses an electronic device.

[0049] Specifically, the electronic device can be a desktop computer, laptop computer, handheld computer, or cloud server, etc. This computer device may include, but is not limited to, a processor and memory. The processor and memory can be connected via a bus or other means. The processor can be a Central Processing Unit (CPU). The processor can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs) or other programmable logic devices, graphics processing units (GPUs), embedded neural network processing units (NPUs) or other dedicated deep learning coprocessors, discrete gate or transistor logic devices, discrete hardware components, or combinations of the above types of chips.

[0050] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs, non-transitory computer-executable programs, and modules. The processor executes various functional applications and data processing by running non-transitory software programs, instructions, and modules stored in memory. Memory may include a program storage area and a data storage area. The program storage area may store the control unit and the application program required for at least one function; the data storage area may store data created by the processor, etc. Furthermore, memory may include high-speed random access memory and non-transitory memory. In some embodiments, memory may optionally include memory remotely located relative to the processor, which can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0051] The present invention also discloses a computer-readable storage medium.

[0052] Specifically, the computer-readable storage medium is used to store a computer program, which, when executed by a processor, implements the methods described above.

[0053] Those skilled in the art will understand that all or part of the processes in the methods described above can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk drive (HDD), or solid-state drive (SSD), etc.; the storage medium can also include combinations of the above types of memory.

Claims

1. A multi-source state trend integrated data fusion method for a weakly supported large-scale renewable energy transmission system, characterized in that, Includes the following steps: Step S1: Identification of multiple data sources and planning of hierarchical access architecture; Step S2: Acquisition, preprocessing, and quality labeling of multi-source state and time series data; Step S3: Time synchronization based on a unified clock and spatial mapping based on topology; Step S4: Construction of multi-scale trend features and extraction of weak support sensitive features; Step S5: Integration of Status and Trend Data and Construction of a Weak Support Risk Indicator System; Step S6: Trend consistency assessment and anomaly detection in weak support scenarios; Step S7: Release of unified data services and interfaces for regulation and market applications; The steps are either executed sequentially or partially parallelized through modular design in engineering applications.

2. The method for integrated data fusion of multi-source state trends in a weakly supported large-scale new energy transmission system according to claim 1, characterized in that, Step S1 is used to systematically sort out the data source types, attributes and importance within the overall architecture of a large-scale new energy transmission system with weak support, and form a hierarchical access architecture plan. (1) Data source classification and modeling Data sources are categorized into three main types: generation-side data, grid-side data, and market and meteorological data. Generation-side data includes SCADA measurements and control commands from wind power, photovoltaic, energy storage, and conventional generating units. Grid-side data includes data from substations, voltage support devices, power flow at transmission sections, PMU and WAMS measurement and protection, and automation operation records. Market and meteorological data includes spot and medium- to long-term electricity price series, clearing volumes, dispatch plans, power forecast curves, and multi-dimensional meteorological elements. For each type of data source, a unified source-level description model is established, defining its sampling period Δt, spatial affiliation, physical dimension, and quality label set Q; (2) Layered access architecture planning The access layer is divided into: edge acquisition layer, backbone aggregation layer and central integration layer. The edge acquisition layer is responsible for establishing a secure channel with field devices and boundary systems. The backbone aggregation layer realizes data aggregation and buffering across voltage levels and scheduling levels. The central integration layer is responsible for unified modeling and services. By specifying the access layer, interface protocol and update strategy for each data source, it is ensured that the key status and trend data have sufficient timeliness and reliability in weak support transmission scenarios.

3. The method for integrated data fusion of multi-source state trends in a weakly supported large-scale new energy transmission system according to claim 1, characterized in that, Step S2 is as follows: Step S2 involves standardized data collection and preprocessing based on the planned data sources, providing quality assurance for subsequent trend construction and risk assessment. (1) Acquisition and buffering mechanism Establish a timestamp-based streaming acquisition channel, sample high-frequency PMU data at the millisecond or tens of millisecond level, sample SCADA and market data at the second to minute level, and achieve reliable data transmission and short-term caching through circular buffer or message queue mechanism; (2) Data cleaning and initial screening of anomalies For the collected time series S i (t) Perform outlier detection, missing value identification, and duplicate value removal operations; (3) Quality markers and pedigree records Add a quality label q to each raw or preprocessed data record i (t)∈Q, and record the data collection time, processing flow identifier and source system identifier to form a traceable data spectrum; Quality labels will be used in subsequent trend feature construction and weak support indicator calculation for weighting or elimination strategies, thereby avoiding low-quality data from misleading risk assessment.

4. The method for integrated data fusion of multi-source state trends in a weakly supported large-scale new energy transmission system according to claim 3, characterized in that, Outlier detection employs the 3σ criterion, box plot method, or threshold determination based on historical probability distribution; missing values ​​are filled in using interpolation between adjacent time points or regression estimation based on adjacent correlation quantities.

5. The integrated data fusion method for multi-source state trends in a weakly supported large-scale new energy transmission system according to claim 1, characterized in that, Step S3 is as follows: In scenarios with weak support for data transmission, the temporal and spatial consistency of multi-source data is crucial for trend analysis and mechanism characterization. Step S3 is to achieve multi-source data alignment under a unified clock and unified topology framework. (1) Unified time base and time scale correction Introducing a unified time base T0, and applying timestamps t from various data sources i To perform correction, for data sources with clock drift or communication delay, the systematic deviation δt is estimated and compensated by calculating the correlation coefficient or the time difference of synchronization events with the reference source. i Thus, the corrected time t is obtained. i =t i δt i ; (2) Multi-scale temporal alignment and resampling For time series with different sampling periods, they will be uniformly mapped to a set of nested time scales {ΔT1,ΔT2,…}. The original sequence will be mapped to different scales that express the state and trend through resampling or aggregation. In the assessment of weak support, the details of rapid changes in voltage and frequency will be maintained at the second level, while the power flow, reserve and market clearing volume will show trends at the minute level. (3) Topology-based spatial mapping and equivalence A network topology model of the power transmission system is constructed, including power generation stations, step-up substations, transmission lines, key receiving-end sections, and receiving-end main grid nodes. Measurement point codes in each data source are mapped to topology node or branch numbers, and data from different systems located at the same electrical node are spatially aligned. For some quantities that cannot be directly mapped, they are mapped to a unified topology region or equivalent node through power flow equivalence or topology folding methods, laying the foundation for the subsequent construction of weak support indicators at multiple levels of nodes-sections-regions.

6. The integrated data fusion method for multi-source state trends in a weakly supported large-scale new energy transmission system according to claim 1, characterized in that, Step S4 is as follows: Step S4, based on the spatiotemporal alignment of state data, constructs multi-scale trend features for weak support mechanisms and mines the feature quantities most sensitive to voltage support and transmission security. (1) Sliding window trend construction At a given time scale ΔT k Below, regarding the key variable x i (t) Construct a sliding window W k (t)=[t L k ,t], calculate the mean μ ik (t), slope g ik (t), volatility σ ik (t) Trend quantity; slope is expressed as: g ik (t)= [x i (t) x i (t L k )] / L k , among which, L k Indicates the duration of the sliding window; Volatility is measured using the standard deviation within a window or the absolute deviation. (2) Multiscale decomposition and frequency band characteristics For high-frequency measurement data of PMU voltage and frequency, discrete wavelet transform or empirical mode decomposition is used to decompose the signal into components of different frequency bands and extract low-frequency trends and high-frequency disturbance characteristics. Low-frequency components are used to characterize slowly changing power flow and voltage evolution, while high-frequency components are used to characterize transient disturbances and oscillations, thus reflecting the multi-timescale dynamic characteristics of weakly supported systems within a unified framework. (3) Weak support sensitive feature extraction Based on sensitivity analysis theory, the sensitivity of node voltage V to injected power ΔP and ΔQ is analyzed. V / P, V / Q. Sensitivity of power flow at key cross-sections to fluctuations in renewable energy output and load. P line / P nen , P line / P loa d performs calculations and constructs a sensitive trend quantity in the trend dimension. S vp (t)= ( V / P) · g p (t), S v _q(t)= ( V / Q) · g_q(t), Where g p (t) and g_q(t) are the trend slopes of the corresponding power injection; S vp (t), S v _q(t) is a sensitive trend indicator for assessing the vulnerability of node voltage; the above sensitive trend characteristics can intuitively reflect the vulnerability of node voltage under the current fluctuation trend.

7. The integrated data fusion method for multi-source state trends in a weakly supported large-scale new energy transmission system according to claim 6, characterized in that, Key variable x i (t) includes key node voltage V, active power flow P, reactive power output Q, and renewable energy output P. nen And reserve margin R.

8. The method for integrated data fusion of multi-source state trends in a weakly supported large-scale new energy transmission system according to claim 1, characterized in that, Step S5 is as follows: Step S5: Based on multi-source and multi-scale trend features, construct a unified state-trend fusion dataset and a weak support risk characterization index to provide a quantitative basis for the assessment and control of outbound security. (1) Construction of unified feature vector of state-trend For each key analysis object, construct an feature vector containing instantaneous state variables and trend variables at a unified time grid point t: z(t) = [V(t), P(t), Q(t), R(t), μ(t), g(t), σ(t), S vp (t), S v _q(t), …] Where V(t), P(t), Q(t), and R(t) represent instantaneous voltage, active power flow, reactive power support, and reserve margin, respectively, and μ(t), g(t), and σ(t) represent trend statistics under different window scales; (2) Design of comprehensive indicators for weak support Based on the eigenvector z(t), a weak support comprehensive index I_w(t) is constructed, using a weighted linear or nonlinear combination: I_w(t) = w1·I_V(t) + w2·I_P(t) + w3·I_Q(t) + w4·I_R(t) + w5·I_trend(t) Where I_V(t) is based on voltage deviation and voltage margin, I_P(t) is based on the cross-sectional power flow ratio, I_Q(t) is based on the utilization level of non-functional capacity, I_R(t) is based on the consumption level of reserve margin, and I_trend(t) is constructed based on trend slope and volatility; weight w i Determined through training with historical data or expert experience; (3) Indicator stratification and alarm threshold I_w(t) is mapped to several risk levels, namely normal, mild weak support, severe weak support, and critical instability, with different color labels and alarm strategies. Differentiated thresholds are set for key nodes or sections to form a multi-level and multi-dimensional weak support risk map, providing a basis for subsequent regulation and market price linkage.

9. The integrated data fusion method for multi-source state trends in a weakly supported large-scale new energy transmission system according to claim 1, characterized in that, Step S6 is as follows: Since the weak support system is sensitive to data quality, a trend consistency assessment and anomaly detection mechanism is introduced in step S6 to ensure the reliability of the state-trend fusion data. (1) Physical consistency verification Based on the power flow balance constraint and energy conservation relationship, the consistency of multi-source state variables is checked to examine the balance relationship between station output, plant power, external power transmission and network loss; whether the changes in node voltage and reactive power output conform to the static characteristic curve; if the observed trend violates the physical constraints, the corresponding data segment is marked as abnormal and its weight in the calculation of weak support index is reduced. (2) Correlation analysis between trends By calculating the correlation coefficients or mutual information between the trends of different variables, namely the trend of new energy output and the trend of cross-sectional power flow, the trend of voltage and the trend of reactive power output, and the trend of electricity price and the trend of power flow, abnormal combinations inconsistent with the weak support mechanism can be identified. (3) Anomaly detection and automatic handling Anomaly detection is performed on the state-trend feature vector z(t) using methods based on statistical thresholds, cluster distance, or simple machine learning models. Data segments identified as anomalous are automatically processed, i.e., marked, removed, or reconstructed by interpolating data from adjacent time periods and neighboring nodes, while retaining the original records for post-analysis and auditing.

10. The method for integrated data fusion of multi-source state trends in a weakly supported large-scale new energy transmission system according to claim 1, characterized in that, Step S7 is as follows: Based on the above access and integration steps, step S7 constructs a unified data service layer to provide standardized interfaces for external security assessment, collaborative regulation and market linkage. (1) Unified data view and topic view A unified data view is constructed around the hierarchical structure of nodes, sections, channels, regions, and markets, providing a comprehensive display of the status, trends, and weak support indicators for key objects; Simultaneously, we construct a security assessment view, a new energy fluctuation view, a channel margin view, and a price-trend linkage view to meet the needs of different applications; (2) Standardized interface and subscription mechanism The design incorporates standardized query and subscription interfaces, allowing external applications to subscribe to the required data based on time intervals, spatial ranges, and metric types. The data service layer supports both real-time push and historical playback modes, while retaining access logs and call records to meet security auditing and performance optimization requirements. (3) Coupling method between control and market decision-making modules Without altering the main control logic of the existing scheduling and control system, weak support indicators and trend information are embedded as constraints or penalties into the output-pricing decision model for power transmission control, reactive power optimization, and participation in the spot and ancillary service markets. This achieves a systematic linkage between enhanced perception on the data side, explicit constraints on the model side, and closed-loop execution on the control side.

Images