A power data risk early warning method and system

Abstract

This invention discloses a power data risk early warning method and system, relating to the field of power system monitoring and intelligent early warning technology. The method includes the following steps: collecting multi-source operational data throughout the entire power system operation process; performing time synchronization processing on data from different sources; establishing a unified time scale and generating a rhythm reference; and constructing a dynamic flow velocity map for queuing analysis. This invention achieves continuous expression of data flow through multi-source data time synchronization and dynamic flow velocity modeling, establishing a unified time reference, and identifying buffer delays and backlog trends in real time to avoid monitoring lag. Through sequence detection, out-of-order point identification, and dynamic rhythm adjustment, it achieves time-series reorganization of old and new data, outputting risk early warning information in advance, and improving the real-time performance and prevention efficiency of power system risk identification.

Description

Technical Field

[0001] This invention relates to the field of power system monitoring and intelligent early warning technology, specifically to a power data risk early warning method and system. Background Technology

[0002] Power data risk early warning refers to a proactive prevention and control technology that utilizes the analysis, modeling, and intelligent identification of large amounts of operational data collected from various stages of power system operation, including transmission, distribution, and consumption, to detect potential operational anomalies or safety hazards in advance. This technology continuously collects and synchronizes multi-source data such as voltage, current, frequency, load, power factor, and equipment temperature to construct dynamic time-series characteristics reflecting changes in system state. Combined with machine learning, pattern recognition, or trend analysis algorithms, it identifies and evaluates signals of abnormal fluctuations, sudden changes, or deviations from established patterns, thereby issuing early warning information before risks spread or faults occur. In this way, power data risk early warning can provide proactive decision support for dispatching operations, equipment maintenance, and safety management, promoting the transformation of the power system from reactive response to proactive prevention, and achieving intelligent, visualized, and safe management of the operation process.

[0003] The existing technology has the following shortcomings: In existing technologies, when power systems experience severe load fluctuations, monitoring terminals often need to collect multi-source operational data at high frequency and transmit it to the central analysis unit in real time. However, due to the rapid increase in data flow within a short period, the cache management mechanism is prone to processing queue disorder, leading to a backlog of monitoring data in the cache and delaying the entry of newly collected data packets into the processing stage. At this time, the system is still parsing operational data from older periods, and the latest risk signals fail to be output in a timely manner, resulting in delayed early warning information. If this lag persists, a significant time misalignment will occur between the power grid's operational status and early warning feedback. Management personnel will be unable to respond effectively based on the actual risk situation during critical periods, potentially missing emergency response windows for sudden load increases, equipment overheating, or voltage anomalies, increasing the potential risks of system instability and equipment damage.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a power data risk early warning method and system to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a power data risk early warning method, comprising the following steps: Collect multi-source operation data throughout the entire power system operation process, perform time synchronization processing on data from different sources, establish a unified time scale and generate a rhythm reference, and construct a dynamic flow rate map for queuing analysis; Based on the dynamic flow rate graph, time-dimensional queuing analysis is performed on the running data in the cache queue to depict the queuing path of the data in the cache, capture cache latency fluctuations, and generate reverse backlog candidate sequences to provide a basis for anomaly localization. Sequence detection is performed on the reverse backlog candidate sequence to identify time sequence aberration segments, extract trigger rhythm markers, and form a list of out-of-order points to reveal abnormal processing order of cached data; By combining the list of misordered points, we conduct a retrospective analysis of the operational data trajectory before and after load changes, extract the delay patterns at the time of cache switching, identify key misordered windows, and establish the response priority relationship for data processing. Dynamic adjustments are performed within the critical misalignment window. Newly acquired data segments are directly transmitted, while old data segments in the buffer are read back with a delay. Data timing is reassembled through dynamic rhythm control of reversal, brief stop, and release. The reassembled real-time operating status characteristics are output, and power data risk warning information is generated based on these characteristics. Warning results are output before the operational risks are formed.

[0007] Preferably, the steps for generating the dynamic velocity map are as follows: Collect multi-source operation data throughout the entire power system operation process, including line voltage, current, power factor and frequency data in the transmission link, load power, node voltage, current waveform and phase angle data in the distribution link, and equipment temperature, line temperature, energy consumption rate and power quality index data in the power consumption link, and generate a timestamp for each piece of operation data. The collected multi-source operational data is time-synchronized according to timestamps, and the data from different sampling channels are rearranged and aligned under a global time scale. A rhythm reference is established under a unified time scale, and a comprehensive rhythm curve is generated by calculating the rate of change of operating parameters at adjacent time nodes; A dynamic flow rate map is generated with a unified time scale as the horizontal axis and the data change rate and rhythm change trend as the vertical axis. The change trajectory of multi-source running data is plotted on the time axis and a continuous time distribution is formed for queuing analysis.

[0008] Preferably, the steps for generating the reversed back-compression candidate sequence are as follows: Based on the time scale information and data flow rate characteristics of the dynamic flow rate map, the running data in the cache queue is sorted by time, and the running data is arranged according to the sampling time label to form a time-ordered cache queue sequence; Based on the time-ordered cache queuing sequence, establish the path mapping relationship between running data in time and cache space, and draw the time trajectory of running data from entering the cache, staying and being transmitted to the processing end; Detect changes in time intervals in the path trajectory, identify the locations of sudden increases in time intervals between adjacent running data, and determine the delay fluctuation segments; Centered on the delayed fluctuation section, the stagnant running data units are arranged in reverse time order to generate a reverse backlog candidate sequence, and the sequence is mapped on the dynamic flow map to determine the anomaly location.

[0009] Preferably, during the generation of the reverse backlog candidate sequence, the start time, duration and running data index of the delay fluctuation segment are identified, and the running data within the delay fluctuation segment are arranged in descending order of time. At the same time, the time position of the reverse backlog candidate sequence is mapped to the dynamic flow rate diagram to form the positioning interval of the delay fluctuation on the time axis, which is used to accurately calibrate the time range of cached data backlog.

[0010] Preferably, the steps for forming the list of out-of-order points are as follows: Obtain the reverse backlog candidate sequence, organize the time attributes of the running data records, arrange them according to the sampling time order to form a time index, and calculate the difference between the sampling time and the release time to generate the cache dwell time sequence; After the time attributes are sorted out, the reverse backlog candidate sequence is sequentially detected. The relationship between the sampling time and the release time is compared, and the data segment whose release time is earlier than the previous item corresponding to the sampling order is identified and the time sequence abnormal segment is determined. Extract rhythm change nodes from the identified time sequence anomalies, and generate trigger rhythm markers based on the turning points of the time difference change direction. After extracting the trigger rhythm markers, the rhythm markers are bound to the abnormal segment indexes and arranged in chronological order to form a list of out-of-order points, which is used to reveal abnormalities in the processing order of cached data.

[0011] Preferably, during the process of generating the list of out-of-order points, the sampling time, release time, cache dwell time, abnormal segment number, rhythm conversion direction and associated data index of each trigger rhythm marker are recorded, and a time out-of-order index table is established on the time axis to ensure the continuity and traceability of out-of-order points in the time dimension, thereby accurately revealing the distribution pattern of cache data order anomalies.

[0012] The preferred method for determining the key misaligned window is as follows: Obtain a list of out-of-order points. Using the time marker of the out-of-order point as the center, backtrack to extract the running data segment before the out-of-order point occurred and backtrack to extract the running data segment after the out-of-order point occurred. Then, stitch the two sets of running data together under a unified time scale to form a continuous running data trajectory. By comparing and analyzing the spliced ​​running data trajectory, the latency characteristics of the cache switching moment are determined by comparing the changing trends of running data before and after the misorder point, and the cache switching latency time is generated. Establish a correspondence between cache switching latency and time location, and determine the key misalignment window based on the latency distribution; Within the critical misalignment window, a response priority relationship for data processing is established based on the magnitude of changes in operating parameters and the time position, and a response priority relationship table is formed.

[0013] Preferably, the following steps are taken: Dynamic adjustment is performed within the critical misalignment window; newly acquired data segments are directly transmitted; old data segments in the buffer are read back with delay; data timing reassembly is completed through rhythm control of reversal, brief pauses, and re-release; and the reassembled real-time operating status characteristics are output, generating power data risk warning information. After the key misalignment window is determined, the data within the window is extracted and classified, and the newly acquired data fragments in real time are distinguished from the old data fragments in the cache that have not been released and a time correspondence is established. After data classification and alignment are completed, rhythmic transmission adjustment is performed on the data within the key misalignment window. Newly acquired data segments are transmitted directly, while old data segments in the buffer are released with a delay and output alternately according to the sampling time order. During the dynamic rhythm adjustment process, sequential control operations such as reversal, brief stop and release are performed on the data stream to make the data continuously arranged on the time axis to form a time chain; After the data stream time sequence is reassembled, the reassembled data is output in chronological order to generate real-time operating status characteristics, and power data risk warning information is generated based on the operating status characteristics.

[0014] A power data risk early warning system includes a data acquisition and modeling module, a cache analysis module, an out-of-order identification module, a delay decision-making module, and a regulation early warning module. The data acquisition and modeling module collects multi-source operational data throughout the entire power system operation process, performs time synchronization processing on data from different sources, establishes a unified time scale and generates a rhythm reference, and constructs a dynamic flow rate map for queuing analysis. The cache analysis module performs time-dimensional queuing analysis on the running data in the cache queue based on the dynamic flow rate graph, depicts the queuing path of data in the cache, captures cache latency fluctuations, and generates a reverse backlog candidate sequence to provide a basis for anomaly localization. The out-of-order detection module performs sequence detection on the reverse backlog candidate sequence, identifies abnormal time sequence segments, extracts trigger rhythm markers, and forms a list of out-of-order points to reveal abnormal processing order of cached data. The delay decision module, in conjunction with the list of misordered points, performs backtracking analysis on the operational data trajectory before and after load changes, extracts the delay patterns at the time of cache switching, identifies key misordered windows, and establishes the response priority relationship for data processing. The adjustment and early warning module performs dynamic adjustment within the critical misalignment window, performs direct transmission of newly acquired data segments, performs delayed readback of old data segments in the cache, and reassembles the data timing through dynamic rhythm control of reversal, brief stop and re-release, outputs the real-time operating status characteristics after reassembly, and generates power data risk early warning information based on the characteristics, outputting the early warning result before the operation risk is formed.

[0015] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention establishes a dynamic flow velocity diagram by uniformly synchronizing and modeling the rhythm of multi-source operational data over time, thereby continuously expressing the data flow patterns. This ensures consistency and traceability of data from different sources during power system operation under the same time reference. By depicting cache queuing paths and capturing delay fluctuations, abnormal rhythms in data flow during transmission and processing can be identified in real time, allowing for early detection of data backlog trends. This effectively avoids monitoring lag caused by cache queue disorder, keeping system operation status and data analysis results synchronized and improving the real-time accuracy of risk identification.

[0016] This invention extracts cache switching delay patterns and establishes response priority relationships by sequentially detecting reverse backlog candidate sequences and generating a list of misordered points, combined with data trajectory backtracking before and after load changes. This enables the system to automatically adjust its processing pace based on data importance and time urgency within critical misalignment windows. Through dynamic adjustment methods such as reversal, brief pauses, and re-release, the system achieves temporal reorganization of old and newly acquired data in the cache, ensuring that risk signals are identified and output promptly before they spread, thereby improving the early warning capability and decision support ability of power data risks. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0018] Figure 1 This is a flowchart of a power data risk early warning method according to the present invention.

[0019] Figure 2 This is a schematic diagram of a power data risk early warning system according to the present invention. Detailed Implementation

[0020] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0021] This invention provides, for example Figure 1 The power data risk early warning method shown includes the following steps: Collect multi-source operation data throughout the entire power system operation process, perform time synchronization processing on data from different sources, establish a unified time scale and generate a rhythm reference, and construct a dynamic flow rate map for queuing analysis; To acquire multi-source operational data covering the entire power system operation process, and establish a unified time scale and rhythm reference in the time dimension, enabling various operational data to form a dynamic flow rate diagram that can be used for subsequent queuing analysis, the specific implementation steps are as follows: Multi-source data acquisition is conducted for characteristic data from different stages of power system operation. This includes line voltage, current, power factor, and frequency data for transmission; load power, node voltage, current waveforms, and phase angle data for distribution; and equipment temperature, line temperature, energy consumption rate, and power quality indicators for consumption. Each type of operational data is acquired by independent acquisition devices within fixed sampling intervals, generating timestamps based on a high-precision time source to record the data acquisition time. To prevent time drift between acquisition channels during multi-source acquisition, the sampling start time of all acquisition channels is synchronized and adjusted at the beginning of each data acquisition cycle, and the sampling end time is recorded after each sampling is completed. This method ensures the temporal continuity and sampling consistency of data from transmission, distribution, and consumption stages. After acquisition, all operational data is stored in a buffer queue in chronological order of sampling time. Each data record simultaneously includes the measured value, timestamp, acquisition source identifier, and corresponding operational stage label, thus forming a multi-source synchronous sampling dataset.

[0022] Time synchronization processing is performed on the acquired multi-source operational data to eliminate deviations in sampling period and sampling time between different sampling channels. The time synchronization process uses a global time scale as a reference, rearranging and aligning various operational data according to timestamps. For channel data with high sampling frequency, time nodes are re-divided on the time axis using the smallest time interval as the unit, mapping the original sampled data to a unified time series. For channel data with relatively low sampling frequency, transition time nodes are generated between adjacent data using time interpolation methods to maintain continuity under a unified time scale. After time synchronization processing, the multi-source operational data form a one-to-one correspondence on the same time axis, simultaneously containing multiple operational characteristics such as voltage, current, power factor, frequency, load power, equipment temperature, and line temperature at each time node. Through the establishment of a unified time scale, all operational parameters can change synchronously in the time dimension, thus providing a complete and continuous time series input for the subsequent establishment of rhythm references.

[0023] After time synchronization is completed, a rhythm reference is established for the multi-source operational data sequence. The rhythm reference establishment process is based on a time series under a unified time scale. By quantitatively analyzing the changes in operational data between consecutive time nodes, a time characteristic curve characterizing the rhythm of power system operational state changes is formed. Specifically, the rate of change of voltage, current, load power, and equipment temperature between adjacent time nodes is calculated sequentially, and continuous change curves are plotted on the time axis according to the rate of change and time interval. Then, based on these change curves, data reflecting the changing trend of system operational rhythm in the time dimension are extracted, and these trends are combined and mapped to form a comprehensive rhythm curve. This comprehensive rhythm curve reflects the data update rate, state fluctuation frequency, and rhythmic characteristics of parameter changes during power system operation, constituting a rhythm reference under a unified time scale. The establishment of the rhythm reference ensures the synchronicity and rhythmic consistency of operational data from different sources during the time-varying process, providing a rhythm baseline for the construction of dynamic velocity maps.

[0024] After establishing the rhythm reference, a dynamic velocity map is generated by mapping the continuous temporal distribution of multi-source operational data using a unified time scale as the horizontal axis and the data change rate and rhythm change trend as the vertical axis. During the construction process, the rhythm curve obtained in the previous step is divided into multiple time segments, and the multi-source operational data corresponding to each segment are arranged chronologically on the time axis. For operational parameters such as voltage, current, frequency, and load power, data change trajectory curves are plotted separately within each time segment, and the curves of different parameters are superimposed with the same time axis as a reference to form a multi-dimensional time distribution map. Subsequently, the change rates of various operational parameters within the same time segment are merged to form a single temporal velocity curve. This velocity curve continuously represents the flow characteristics of operational data in the time dimension, reflecting the rhythm of data updates and the speed of system state changes. In the dynamic velocity map, each time segment contains synchronized multi-source operational data and its change trend, and the entire velocity map forms a continuous temporal distribution from the starting point to the current moment. By analyzing the dynamic flow rate diagram, the rhythm characteristics and data flow patterns of the power system at different operating stages can be identified, providing continuous time basis for subsequent buffer queuing analysis, delay identification, and data processing priority determination.

[0025] Based on the dynamic flow rate graph, time-dimensional queuing analysis is performed on the running data in the cache queue to depict the queuing path of the data in the cache, capture cache latency fluctuations, and generate reverse backlog candidate sequences to provide a basis for anomaly localization. Based on the dynamic flow rate map, the running data in the cache queue is subjected to time-dimensional queuing analysis, and the data flow path in the cache is depicted to capture latency fluctuations, thereby generating a reverse backlog candidate sequence and providing a basis for anomaly localization. The specific implementation steps are as follows: After obtaining the dynamic flow rate map, the temporal distribution of the running data in the buffer queue is organized based on the time scale information and data flow rate characteristics in the dynamic flow rate map. The buffer queue contains running data transmitted from different acquisition stages of the power system. This data is temporarily stored in the buffer space according to the acquisition sequence, awaiting parsing by the central processing unit. Due to the different transmission delays of different acquisition paths, the order of various data in the buffer queue is not completely consistent with the actual sampling time. Therefore, at this stage, the unified time scale of the dynamic flow rate map is used as a benchmark to sort the running data in the buffer according to its sampling time label. During the sorting process, the time label of each running data is compared with the corresponding time node in the dynamic flow rate map and arranged in ascending order of time, forming an initial temporally ordered buffer queue sequence. In this process, the time position of each data record is mapped so that it has a clear coordinate position on the time axis of the dynamic flow rate map. In this way, the running data in the buffer is rearranged in chronological order, forming a basic queue structure that reflects the actual sampling time sequence, providing a reference for subsequent path depiction.

[0026] After time sorting, the flow trajectory of running data in the cache queue is continuously depicted along the time dimension, based on the sorted cache queuing sequence. This process uses the time change trend in the dynamic flow graph as a reference, mapping the time coordinate of each data item in the cache to its cache storage location, ensuring that each data unit has a traceable path identifier in both time and space dimensions. By sequentially traversing the data records in the cache, a mapping relationship of the time flow path is established, forming a coherent trajectory on the time axis from data entering the cache, remaining there, to being transmitted to the processing end. To ensure the continuity of the trajectory, a time interval relationship is established between every two adjacent data items. By calculating the storage order of adjacent data in the cache, a queuing path line of data in the cache space is drawn. This path line corresponds to the change trajectory of the cache queue along the time dimension, reflecting the flow rhythm and occupancy order of data in the cache. By continuously depicting the path line, the accumulation and release patterns of data in the cache can be observed in different time periods, thus providing a basis for subsequently capturing latency fluctuations.

[0027] After the cache queuing path is drawn, anomalies in the time intervals along the path are identified to capture cache latency fluctuations. Latency fluctuations typically manifest as a disruption of the smooth continuity of the cache path on the time axis, or a sudden increase in the time interval between adjacent data. Therefore, interval detection is performed on each segment of the path along the time dimension, calculating the change in the time interval between adjacent data. When several consecutive data points in the cache exhibit a continuously increasing time interval, it indicates that data congestion or latency backlog has occurred in that cache segment. Simultaneously, the trend of cache queuing speed changes can be determined by comparing the path's tilt angle over different time periods. A sudden change in the path's tilt indicates a change in the data flow rate in the cache, reflecting latency fluctuations in the cache management process. Through the analysis of these time interval anomalies and path changes, the time intervals and corresponding data ranges of latency fluctuations in the cache can be accurately determined, thus revealing the distribution pattern of cache latency.

[0028] After identifying delay fluctuations, aggregate analysis is performed on the affected segments to generate reverse backlog candidate sequences. This process centers on the time nodes of the delay fluctuation segments, summarizing data fragments before and after the delay occurs, and arranging the data units remaining in the delay fluctuation segments in reverse chronological order to form reverse backlog candidate sequences. During the generation of reverse backlog candidate sequences, the start time, duration, and corresponding data index of the delay fluctuations are identified, ensuring that each candidate sequence corresponds to a specific cache time segment and data range. Simultaneously, the positions of these sequences on the time axis are mapped back to the dynamic velocity graph, making them appear as localized velocity decreases or stagnation areas on the velocity change curve. This mapping method allows for intuitive observation of the temporal location and data distribution characteristics of delay fluctuations in the cache, providing direct evidence for anomaly localization. The reverse backlog candidate sequences not only record the segments where delay fluctuations occur in the cache but also reflect the order and duration of delay formation, serving as a crucial foundation for subsequent time-disorder identification and anomaly tracking.

[0029] Sequence detection is performed on the reverse backlog candidate sequence to identify time sequence aberration segments, extract trigger rhythm markers, and form a list of out-of-order points to reveal abnormal processing order of cached data; To ensure that the data order in the reverse backlog candidate sequence can be fully detected and that out-of-order segments are identified in the time dimension, thereby extracting trigger rhythm markers that reflect changes in cache processing rhythm and forming a list of out-of-order points to reveal abnormalities in the cache data processing order, the specific implementation steps are as follows: After obtaining the reverse backlog candidate sequence, the time attributes of each running data record in the sequence are organized to make all data comparable in the time dimension. The reverse backlog candidate sequence consists of data segments with latency fluctuations in the cache. Each data record contains the sampling time, the time of entering the cache, the time spent in the cache, and the time of being released from the cache. To establish the basis for time sequence detection, all data are first sorted according to the data sampling time, with data with earlier sampling times arranged first and data with later sampling times arranged last, and each data is assigned a consecutive time index number. After sorting, the correspondence between the sampling time and release time of each data is recorded, and the time difference between the two is calculated to form a cache dwell time sequence. Subsequently, all data are rearranged according to the order of cache release time to form a cache output order sequence. By comparing the differences between the sampling order and the output order, a reference table for time sequence analysis is constructed. This reference table records the sampling time, release time, cache dwell time, and time index position of each data, providing basic data support for the next step of order detection.

[0030] After the time attributes are processed, the reverse backlog candidate sequence is subjected to time order detection to identify data segments with disordered time sequences. The core of the order detection is to determine the logical consistency between the sampling time and the release time of the data. In the specific operation, starting from the data with the smallest time index number, the sampling time and release time of two adjacent data are compared sequentially. When the release time of the later data is earlier than the release time of the earlier data, it indicates that the cache processing order has reversed. To ensure the continuity and completeness of the detection process, after each set of data comparison, the two data with reverse order relationship and their adjacent data are further detected to determine whether a continuous reverse order segment is formed. When three or more consecutive data have a release time earlier than the preceding item in their sampling order, the segment is defined as a time order abnormal segment. During the identification process, the start position, end position, and number of data contained in each abnormal segment are recorded, and the index range of the abnormal segment in the reverse backlog candidate sequence is marked. After the detection is completed, the entire reverse backlog candidate sequence is divided into two types: normal order segments and abnormal order segments, providing a time partitioning basis for the subsequent extraction of trigger rhythm markers.

[0031] After identifying time-series aberration segments, rhythmic feature analysis is performed on the boundary positions of each aberration segment to extract trigger rhythm markers. The purpose of these trigger rhythm markers is to identify the specific time points at which cached data transitions from normal processing order to aberration order or recovers from anberration order to normal order. In the specific implementation, the sampling and release times of several data points before and after the boundary of the time-series aberration segment are compared and analyzed. When a turning point is found where the direction of the time difference changes from increasing to decreasing or vice versa, this time point is defined as a rhythm change node, and the node's sampling time, release time, cache dwell time, and corresponding index number are recorded to generate trigger rhythm markers. Each trigger rhythm marker represents the moment of a sequence transition in the cache, reflecting the start or end point of the data flow's rhythm change in the cache queue. By arranging all trigger rhythm markers in chronological order, a time series reflecting the cache rhythm switching pattern can be formed, revealing the dynamic process of data transitioning from continuous order to reverse backlog and then back to normal during cache processing.

[0032] After extracting the trigger rhythm markers, all rhythm markers and time-sequence anomaly segments are organized and summarized to form a list of misordered points. The generation process of the misordered point list includes three steps: First, each trigger rhythm marker is bound to its corresponding anomaly segment index to ensure that each misordered point has its corresponding anomaly segment background; second, all misordered points are rearranged in chronological order to establish a time misordered index table, making the misordered points traceable on the timeline; finally, descriptive information is added to each misordered point, including sampling time, release time, cache dwell time, anomaly segment number, rhythm transition direction, and associated data index. In this way, the misordered point list not only records the occurrence time of time-sequence misordering phenomena but also reveals the rhythmic change pattern of cached data from normal to abnormal and then back to recovery over time. The generated misordered point list can fully reflect the abnormal distribution of cached data processing order, providing a clear time reference and anomaly identification basis for subsequent latency pattern analysis, key misalignment window location, and data adjustment decisions.

[0033] By combining the list of misordered points, we conduct a retrospective analysis of the operational data trajectory before and after load changes, extract the delay patterns at the time of cache switching, identify key misordered windows, and establish the response priority relationship for data processing. By combining the list of misordered points, a retrospective analysis of the operational data trajectory before and after load changes is performed to extract the latency patterns at cache switching times, and further determine the key misordered windows to establish the response priority relationship for data processing. The specific implementation steps are as follows: After obtaining the list of out-of-order points, the operational data trajectory is traced back to the corresponding time segment using each out-of-order point as the basis for time positioning. The list of out-of-order points contains detailed information such as sampling time, release time, buffer dwell time, and rhythm conversion direction. Each out-of-order point corresponds to the location where the sequence anomaly occurred during buffer processing. To clarify the relationship between out-of-order points and the operating state of the power system, at this stage, a certain time interval is traced back from the time identifier of each out-of-order point to extract the operational data segment before the out-of-order point occurred; simultaneously, the same time interval is traced backward to extract the operational data segment after the out-of-order point occurred. The two time segments represent the operating state of the power system before and after the buffer anomaly occurred. The extracted operational data includes voltage fluctuations, current changes, load power curves, frequency offsets, and equipment temperature changes, all of which come from the operational data sequence after time synchronization processing. After completing the data extraction, the data from the two time segments are spliced ​​into a continuous time series according to a unified time scale, and the precise location of the out-of-order point is marked on the time axis. In this way, the trajectory of the power system's operating state before and after the misalignment point can be observed within the same time frame, providing a time basis for the subsequent extraction of delay patterns.

[0034] After obtaining the operational data trajectories before and after the misordering point, the operational data of the two time periods are compared and analyzed to identify the latency characteristics during cache switching. Latency characteristics are extracted primarily by comparing the trends of operational data changes before and after the misordering point. In practice, the operational data before the misordering point is considered the response curve under normal cache conditions, and the operational data after the misordering point is considered the response curve under abnormal cache conditions. By comparing the rate of change, response start time, and fluctuation amplitude of the two curves at the same time scale, the latency phenomenon exhibited at the cache switching point can be determined. The existence of latency is usually manifested as a time lag in operational data after the misordering point; that is, the load change has already occurred in actual operation, while the data in the cache remains in the old state. To accurately depict the latency pattern, the correspondence between the data before and after the misordering point on the time axis is compared, and the time interval between the actual occurrence of the load change and the cache response start time is calculated. This time interval is the cache switching latency. By performing comparative analysis on data before and after multiple misordering points in the same way, the distribution characteristics of cache switching latency in different operating periods can be obtained, forming a set of latency pattern data reflecting the dynamic response of the cache.

[0035] After extracting the patterns of cache switching delays, time aggregation analysis is performed on the delay pattern data to identify critical misalignment windows. A critical misalignment window refers to the time interval where cache delay reaches its peak or the rate of delay change is most pronounced, reflecting the concentrated manifestation of system response lag during cache switching. In practice, a correspondence is established between the cache switching delay time corresponding to each misalignment point and its time position, and a delay time distribution curve is plotted on the time axis. Subsequently, based on the continuous fluctuations of the delay curve, the trend of delay time changes is segmented for analysis. When the delay time continuously increases and remains high within a certain time interval, this interval is defined as the critical misalignment window. Within the critical misalignment window, there is a strong time misalignment between cached data and the actual operating state, and the system's data parsing and response commands may be asynchronous during this period. By identifying the critical misalignment window, the most sensitive period for cache processing in power system operation can be identified, providing a time range for subsequent response priority decisions.

[0036] After identifying the critical misalignment window, a data processing priority relationship is established to ensure that data with a significant impact on operational risks is processed first when delays occur. The establishment of this priority relationship is based on the type of operational data and its time urgency within the critical misalignment window. First, based on the magnitude and duration of changes in operational parameters within the critical misalignment window, operational data such as voltage changes, current fluctuations, load surges, equipment temperature increases, and frequency deviations are categorized, with parameters directly affecting power system stability assigned high priority. Then, based on the temporal position of each high-priority parameter within the critical misalignment window, the data response order is determined, giving higher priority to data that is closer in time to the actual operational abrupt change point. Finally, the priorities of various operational data types are combined with cache scheduling rules to establish a response priority relationship table based on time and importance. This table clearly defines the processing order and response sequence of various operational data types when cache switching delays occur, thereby achieving timely feedback on critical operational states.

[0037] Dynamic adjustment is performed within the critical misalignment window. Newly acquired data segments are directly transmitted, while old data segments in the buffer are read back with a delay. Data timing is reassembled through dynamic rhythm control of reversal, brief stop and release. The reassembled real-time operating status characteristics are output, and power data risk warning information is generated based on these characteristics. Warning results are output before the operating risk is formed. Within the critical misalignment window, dynamic adjustment of operational data is performed to rearrange newly acquired data segments with lagging old data segments in the buffer along the time dimension, maintaining the continuity of the processing rhythm. This achieves data time sequence reorganization, and power data risk early warning information is generated based on the reorganized operational status characteristics to output early warning results before operational risks materialize. The specific implementation steps are as follows: After identifying the critical misalignment window, the data within this time interval is extracted and classified. Real-time newly acquired data segments are distinguished from older data segments that have not yet been released from the cache, and a time correspondence is established. The critical misalignment window is a time interval identified during the delay pattern analysis phase, representing the period of concentrated cache delay. In this phase, operational data segments generated within this window are first obtained from the data acquisition channel, including transmission line voltage change data, distribution circuit current change data, user-side load power fluctuation data, equipment temperature change data, and frequency fluctuation data. Each operational data segment includes the sampling time, acquisition stage identifier, and corresponding status value. Simultaneously, older data segments that have not yet entered the processing stage within the same time window are extracted from the cache queue; these data segments are held in the cache due to processing delays. Subsequently, the new and old data segments are aligned based on the sampling time to ensure that both types of data have a common reference coordinate on the time axis. This alignment process clarifies the temporal relationship between the newly acquired and older data segments, providing a time-sequencing basis for subsequent dynamic adjustment steps.

[0038] After data classification and alignment, rhythmic transmission adjustments are performed on the operational data within key misalignment windows to ensure that the latest data fragments are prioritized for transmission and timely participation in status analysis, while old data in the cache is released sequentially according to time. At this stage, a continuous data transmission channel is first established, directly introducing newly acquired data fragments into the data processing flow. This ensures that this data can be transmitted directly to the real-time parsing stage without going through the cache, thus maintaining the timeliness of system status data. Simultaneously, a delayed release operation is performed on old data fragments remaining in the cache. The specific process of delayed release involves reading old data in the cache in segments according to the sampling time order, with the length of each read segment consistent with the cache release rate to avoid time overlap caused by simultaneous data release. To prevent sudden concentration of data flow during release, cache release is periodically advanced according to a set time rhythm, alternating between the reading of old data and the transmission of new data, thus forming a continuous and conflict-free data flow path on the timeline. In this way, real-time data can be processed first, while old data in the cache is read back at a controlled pace, achieving coordinated allocation of time and processing resources.

[0039] During dynamic rhythm adjustment, to further eliminate timing discrepancies caused by cache latency, sequential control operations involving reversal, brief pause, and re-release are performed on the data stream within critical misalignment windows, thereby restoring the continuous arrangement of running data in the time dimension. The reversal operation involves identifying old data segments in the cache with reversed time order and readjusting their arrangement, rearranging these lagging segments according to their sampling time sequence to the correct time interval, allowing the data stream to once again present a time progression from the past to the present. The brief pause operation is an instantaneous control measure implemented when the rate of cache release and new data transmission is inconsistent. By temporarily halting the direct transmission of new data, a time interval is created for cache release, allowing old data to be read back and the cache space to be released during this interval, thus preventing data stream overlap or overwriting in time. The re-release operation is executed immediately after the brief pause, re-entering the transmission channel with the new data temporarily stored during the pause, restoring the continuous time progression rhythm of the data stream. Through continuous coordination of reversal, brief pause and re-release, all data streams within the critical misalignment window are rearranged on the timeline, and the old and new data are restored to a unified time chain in logical order, thus unifying the timing of cache processing and real-time acquisition.

[0040] After dynamic adjustment and time-series reassembly of the data stream, the reassembled data is output in chronological order to generate real-time operating status characteristics of the power system. Based on these characteristics, risk warning information is generated, thus providing early warning results before operational risks materialize. Specifically, the reassembled data stream is input into the operating status characteristic extraction process, analyzing voltage change trends, current change rates, load power fluctuation amplitudes, equipment temperature rise rates, and frequency offsets at each time node. By comparing the changes in these operating parameters over continuous time periods, characteristic curves reflecting the stability of the power system are obtained. These characteristic curves describe the continuous evolution of the system's operating status over time. Subsequently, by identifying time segments in the characteristic curve where parameter change rates accelerate, fluctuation amplitudes increase, or state changes abruptly, it is determined whether the system's operating status is within a potential risk range. When an abnormal trend is detected in the operating status curve for a certain time period, corresponding power data risk warning information is generated based on the operating characteristics of that time period. The generated warning information includes the risk type, risk occurrence time, involved operating parameters, risk level, and scope of impact. The early warning information is correlated with the reorganized operational data through time tags, so that the operational status and risk information are consistent in time. This allows the early warning results to be output before the risk is formed, providing dispatchers with a basis for early intervention.

[0041] This invention establishes a dynamic flow velocity diagram by uniformly synchronizing and modeling the rhythm of multi-source operational data over time, thereby continuously expressing the data flow patterns. This ensures consistency and traceability of data from different sources during power system operation under the same time reference. By depicting cache queuing paths and capturing delay fluctuations, abnormal rhythms in data flow during transmission and processing can be identified in real time, allowing for early detection of data backlog trends. This effectively avoids monitoring lag caused by cache queue disorder, keeping system operation status and data analysis results synchronized and improving the real-time accuracy of risk identification.

[0042] This invention extracts cache switching delay patterns and establishes response priority relationships by sequentially detecting reverse backlog candidate sequences and generating a list of misordered points, combined with data trajectory backtracking before and after load changes. This enables the system to automatically adjust its processing pace based on data importance and time urgency within critical misalignment windows. Through dynamic adjustment methods such as reversal, brief pauses, and re-release, the system achieves temporal reorganization of old and newly acquired data in the cache, ensuring that risk signals are identified and output promptly before they spread, thereby improving the early warning capability and decision support ability of power data risks.

[0043] This invention provides, for example Figure 2 The power data risk early warning system shown includes a data acquisition and modeling module, a cache analysis module, an out-of-order identification module, a delay decision-making module, and a regulation early warning module. The data acquisition and modeling module collects multi-source operational data throughout the entire power system operation process, performs time synchronization processing on data from different sources, establishes a unified time scale and generates a rhythm reference, and constructs a dynamic flow rate map for queuing analysis. The cache analysis module performs time-dimensional queuing analysis on the running data in the cache queue based on the dynamic flow rate graph, depicts the queuing path of data in the cache, captures cache latency fluctuations, and generates a reverse backlog candidate sequence to provide a basis for anomaly localization. The out-of-order detection module performs sequence detection on the reverse backlog candidate sequence, identifies abnormal time sequence segments, extracts trigger rhythm markers, and forms a list of out-of-order points to reveal abnormal processing order of cached data. The delay decision module, in conjunction with the list of misordered points, performs backtracking analysis on the operational data trajectory before and after load changes, extracts the delay patterns at the time of cache switching, identifies key misordered windows, and establishes the response priority relationship for data processing. The adjustment and early warning module performs dynamic adjustment within the critical misalignment window, performs direct transmission of newly acquired data segments, performs delayed readback of old data segments in the cache, and reassembles the data timing through dynamic rhythm control of reversal, brief stop and re-release, outputs the real-time operating status characteristics after reassembly, and generates power data risk early warning information based on the characteristics, outputting the early warning result before the operation risk is formed.

[0044] The present invention provides a power data risk early warning method, which is implemented through the above-mentioned power data risk early warning system. For details of the specific method and process of the power data risk early warning system, please refer to the above-mentioned embodiment of the power data risk early warning method, which will not be repeated here.

[0045] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A power data risk early warning method, characterized in that, Includes the following steps: Collect multi-source operation data throughout the entire power system operation process, perform time synchronization processing on data from different sources, establish a unified time scale and generate a rhythm reference, and construct a dynamic flow rate map for queuing analysis; Based on the dynamic flow rate graph, time-dimensional queuing analysis is performed on the running data in the cache queue to depict the queuing path of the data in the cache, capture cache latency fluctuations, and generate reverse backlog candidate sequences to provide a basis for anomaly localization. Sequence detection is performed on the reverse backlog candidate sequence to identify time sequence aberration segments, extract trigger rhythm markers, and form a list of out-of-order points to reveal abnormal processing order of cached data; By combining the list of misordered points, we conduct a retrospective analysis of the operational data trajectory before and after load changes, extract the delay patterns at the time of cache switching, identify key misordered windows, and establish the response priority relationship for data processing. Dynamic adjustments are performed within the critical misalignment window. Newly acquired data segments are directly transmitted, while old data segments in the buffer are read back with a delay. Data timing is reassembled through dynamic rhythm control of reversal, brief stop, and release. The reassembled real-time operating status characteristics are output, and power data risk warning information is generated based on these characteristics. Warning results are output before the operational risks are formed.

2. The power data risk early warning method according to claim 1, characterized in that, The steps for generating a dynamic velocity map are as follows: Collect multi-source operation data throughout the entire power system operation process, including line voltage, current, power factor and frequency data in the transmission link, load power, node voltage, current waveform and phase angle data in the distribution link, and equipment temperature, line temperature, energy consumption rate and power quality index data in the power consumption link, and generate a timestamp for each piece of operation data. The collected multi-source operational data is time-synchronized according to timestamps, and the data from different sampling channels are rearranged and aligned under a global time scale. A rhythm reference is established under a unified time scale, and a comprehensive rhythm curve is generated by calculating the rate of change of operating parameters at adjacent time nodes; A dynamic flow rate map is generated with a unified time scale as the horizontal axis and the data change rate and rhythm change trend as the vertical axis. The change trajectory of multi-source running data is plotted on the time axis and a continuous time distribution is formed for queuing analysis.

3. The power data risk early warning method according to claim 2, characterized in that, The steps for generating candidate sequences through reverse accumulation are as follows: Based on the time scale information and data flow rate characteristics of the dynamic flow rate map, the running data in the cache queue is sorted by time, and the running data is arranged according to the sampling time label to form a time-ordered cache queue sequence; Based on the time-ordered cache queuing sequence, establish the path mapping relationship between running data in time and cache space, and draw the time trajectory of running data from entering the cache, staying and being transmitted to the processing end; Detect changes in time intervals in the path trajectory, identify the locations of sudden increases in time intervals between adjacent running data, and determine the delay fluctuation segments; Centered on the delayed fluctuation section, the stagnant running data units are arranged in reverse time order to generate a reverse backlog candidate sequence, and the sequence is mapped on the dynamic flow map to determine the anomaly location.

4. The power data risk early warning method according to claim 3, characterized in that, During the generation of the reverse backlog candidate sequence, the start time, duration, and running data index of the delay fluctuation segment are identified, and the running data within the delay fluctuation segment are arranged in descending order of time. At the same time, the time position of the reverse backlog candidate sequence is mapped to the dynamic flow rate diagram to form the positioning interval of the delay fluctuation on the time axis, which is used to accurately calibrate the time range of cached data backlog.

5. The power data risk early warning method according to claim 3, characterized in that, The steps for forming the list of out-of-order points are as follows: Obtain the reverse backlog candidate sequence, organize the time attributes of the running data records, arrange them according to the sampling time order to form a time index, and calculate the difference between the sampling time and the release time to generate the cache dwell time sequence; After the time attributes are sorted out, the reverse backlog candidate sequence is sequentially detected. The relationship between the sampling time and the release time is compared, and the data segment whose release time is earlier than the previous item corresponding to the sampling order is identified and the time sequence abnormal segment is determined. Extract rhythm change nodes from the identified time sequence anomalies, and generate trigger rhythm markers based on the turning points of the time difference change direction. After extracting the trigger rhythm markers, the rhythm markers are bound to the abnormal segment indexes and arranged in chronological order to form a list of out-of-order points, which is used to reveal abnormalities in the processing order of cached data.

6. The power data risk early warning method according to claim 5, characterized in that, During the process of generating the list of out-of-order points, the sampling time, release time, cache dwell time, abnormal segment number, rhythm conversion direction and associated data index of each trigger rhythm marker are recorded, and a time out-of-order index table is built on the time axis to ensure the continuity and traceability of out-of-order points in the time dimension, thereby accurately revealing the distribution pattern of cache data order anomalies.

7. The power data risk early warning method according to claim 5, characterized in that, The process for determining the key misaligned window is as follows: Obtain a list of out-of-order points. Using the time marker of the out-of-order point as the center, backtrack to extract the running data segment before the out-of-order point occurred and backtrack to extract the running data segment after the out-of-order point occurred. Then, stitch the two sets of running data together under a unified time scale to form a continuous running data trajectory. By comparing and analyzing the spliced ​​running data trajectory, the latency characteristics of the cache switching moment are determined by comparing the changing trends of running data before and after the misorder point, and the cache switching latency time is generated. Establish a correspondence between cache switching latency and time location, and determine the key misalignment window based on the latency distribution; Within the critical misalignment window, a response priority relationship for data processing is established based on the magnitude of changes in operating parameters and the time position, and a response priority relationship table is formed.

8. The power data risk early warning method according to claim 7, characterized in that, The steps for performing dynamic adjustments within a critical misalignment window, directly transmitting newly acquired data segments, delaying the reading back of old data segments in the buffer, and reassembling the data timing sequence through rhythm control of reversal, brief pauses, and re-release, and then outputting the reassembled real-time operating status characteristics and generating power data risk warning information are as follows: After the key misalignment window is determined, the data within the window is extracted and classified, and the newly acquired data fragments in real time are distinguished from the old data fragments in the cache that have not been released and a time correspondence is established. After data classification and alignment are completed, rhythmic transmission adjustment is performed on the data within the key misalignment window. Newly acquired data segments are transmitted directly, while old data segments in the buffer are released with a delay and output alternately according to the sampling time order. During the dynamic rhythm adjustment process, sequential control operations such as reversal, brief stop and release are performed on the data stream to make the data continuously arranged on the time axis to form a time chain; After the data stream time sequence is reassembled, the reassembled data is output in chronological order to generate real-time operating status characteristics, and power data risk warning information is generated based on the operating status characteristics.

9. A power data risk early warning system, used to implement the power data risk early warning method according to any one of claims 1-8, characterized in that, It includes modules for data acquisition and modeling, cache analysis, out-of-order identification, delay decision-making, and adjustment and early warning. The data acquisition and modeling module collects multi-source operational data throughout the entire power system operation process, performs time synchronization processing on data from different sources, establishes a unified time scale and generates a rhythm reference, and constructs a dynamic flow rate map for queuing analysis. The cache analysis module performs time-dimensional queuing analysis on the running data in the cache queue based on the dynamic flow rate graph, depicts the queuing path of data in the cache, captures cache latency fluctuations, and generates a reverse backlog candidate sequence to provide a basis for anomaly localization. The out-of-order detection module performs sequence detection on the reverse backlog candidate sequence, identifies abnormal time sequence segments, extracts trigger rhythm markers, and forms a list of out-of-order points to reveal abnormal processing order of cached data. The delay decision module, in conjunction with the list of misordered points, performs backtracking analysis on the operational data trajectory before and after load changes, extracts the delay patterns at the time of cache switching, identifies key misordered windows, and establishes the response priority relationship for data processing. The adjustment and early warning module performs dynamic adjustment within the critical misalignment window, performs direct transmission of newly acquired data segments, performs delayed readback of old data segments in the cache, and reassembles the data timing through dynamic rhythm control of reversal, brief stop and re-release, outputs the real-time operating status characteristics after reassembly, and generates power data risk early warning information based on the characteristics, outputting the early warning result before the operation risk is formed.

Images