Power plant electric quantity statistical method

Abstract

The invention discloses a power plant electric quantity statistical method, which belongs to the field of electric power system intelligent monitoring and electric quantity statistical, and comprises the following steps: S1, collecting analog quantity signals of current, voltage, frequency and power factor, and carrying out first timestamp correction processing; s2, uploading to an electric quantity main server and a scheduling synchronization node; s3, constructing a minute-level electric quantity data block, performing difference reconstruction on boundary errors of adjacent time periods, and generating a time sequence electric quantity vector in a unified format; s4, performing screening of discontinuous electric quantity fragments on each time sequence electric quantity vector; s5, performing structured analysis on the time sequence electric quantity, extracting a power generation load change section, and generating a partition statistical report; s6, based on a global electric quantity timeline generated by the scheduling synchronization node, performing synchronous verification on a statistical result of the electric quantity main server, and constructing a credible daily statistical electric quantity report; and S7, constructing an edge node topological graph, and associating the edge node topological graph with statistical data. The method has the beneficial effects that the electric quantity precision, continuity and abnormity positioning capability are improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent monitoring and electricity statistics of power systems, and more specifically, to a method for electricity statistics of power plants. Background Art

[0002] With the digitalization of power systems, power plant energy statistics technology has gradually evolved from manual meter reading and centralized SCADA systems to intelligent platforms capable of automated data collection, remote monitoring, and edge processing. In recent years, edge computing and high-precision timing technologies have been gradually introduced into the power industry, significantly improving the real-time performance and data consistency of energy collection. Furthermore, emerging technologies such as 5G communications, hash chain verification, and topology awareness provide richer support for grid operation status analysis.

[0003] However, existing technologies still have the following shortcomings: First, timing mechanisms often rely on a single-source GPS, resulting in poor system anti-interference capabilities and prone to misalignment of power data; second, power transmission paths are single and lack link fault tolerance; third, statistical methods often rely on simple accumulation and averaging, lacking vectorized reconstruction mechanisms, resulting in weak accuracy and structuring capabilities; fourth, anomaly identification relies on set thresholds, resulting in coarse positioning granularity and difficulty in tracing back to the source; and fifth, data consistency checks are often based on time alignment, which cannot guarantee the synchronization of underlying data structures. Therefore, innovative methods such as multi-source timing, dual-channel uploads, hash chain tracking, and topology integration are needed to address the shortcomings of existing statistical methods in terms of high reliability, high accuracy, and traceability. Summary of the Invention

[0004] The purpose of the present invention is to provide a power plant electricity statistics method to solve the problems raised in the above background technology: first, the timing mechanism mostly relies on a single-source GPS, the system has poor anti-interference ability, and easily leads to electricity data dislocation; second, the electricity transmission path is single, and lacks link fault tolerance; third, the statistical methods mostly use simple accumulation and averaging, lack a vectorized reconstruction mechanism, and have weak accuracy and structuring capabilities; fourth, anomaly identification relies on setting thresholds, the positioning granularity is coarse, and it is difficult to trace the source; fifth, data consistency verification is mostly based on time alignment, which cannot guarantee the synchronization of deep data structures.

[0005] Technical solution: A power plant electricity statistics method includes the following steps: S1. Based on the intelligent edge gateway device on the generator side, the analog signals of current, voltage, frequency and power factor are collected and the first timestamp correction is performed; S2. Upload the corrected signal simultaneously to the power master server and the scheduling synchronization node via a dual-channel path; S3. Construct minute-level electricity data blocks on the main power server side, reconstruct the boundary errors of adjacent time periods by difference, and generate a time series electricity vector in a unified format; S4. Perform bidirectional hash tracking processing on each of the time series power vectors to filter out non-continuous power segments; S5. Perform structured analysis on the time series electricity, extract the power generation load change section, and generate a partition statistical report; S6. Based on the global power timeline generated by the scheduling synchronization node, the statistical results of the power master server are synchronized and verified to build a reliable daily power statistics report; S7. Construct an edge node topology map and associate the constructed edge node topology map with statistical data for topological positioning and isolation control of abnormal power segments.

[0006] Preferably, the first timestamp correction process of S1 further includes the following steps: S1-1. The system clock of the generator side smart edge gateway device is aligned with the external timing module through the PPS physical trigger signal, and combined with a temperature-controlled crystal oscillator to stabilize the delayed response; S1-2. Apply the sliding time window mean correction method to fine-tune and compensate the local time offset of the continuously sampled signal to eliminate the timing error caused by dynamic drift.

[0007] Preferably, the S1-1 further comprises the following steps: S1-1-1. Introducing a multi-source redundant timing mechanism that automatically switches to a backup atomic clock signal when the primary clock signal is interrupted. S1-1-2. Set the hard trigger threshold window. The response delay of less than 2 microseconds is directly marked as the high-precision authorization period. S1-1-3. All timing segments are time-traced and labeled to ensure that subsequent abnormal periods can be traced back to the corresponding hardware offset node.

[0008] Preferably, the generation of the minute-level electricity data block of S3 further includes the following steps: S3-1. Normalize the signal data within each 60-second segment and accumulate the raw data segments of equal length; S3-2. Use quintic spline interpolation to generate a smooth transition curve for each edge point, and perform continuity verification based on the curvature of the center segment; S3-3. Encode each piece of processed data into a 128-dimensional vector, retaining the original node index information as a verification field for subsequent chain tracing.

[0009] Preferably, the interpolation reconstruction in S3-2 further includes the following steps: S3-2-1. Select 15 sampling points before and after the boundary within the time window as fitting reference points; S3-2-2. Introduce a curvature control factor to limit the rate of change of the derivative after fitting to be less than the set gradient threshold; S3-2-3. Make secondary adjustments to the interpolation segment based on the actual frequency offset to eliminate the fitting error caused by grid frequency disturbance.

[0010] Preferably, the bidirectional hash tracking process of S4 further includes the following steps: S4-1. Execute SHA-512 algorithm encoding to form a forward linked list structure on the timing power vector; S4-2. Encode the same data in reverse order into a reverse chain and perform hash value verification on each node; S4-3. If the chain continuity is interrupted, mark the interruption point index and isolate the two discontinuous segments before and after.

[0011] Preferably, the synchronization verification process of S6 further includes the following steps: S6-1. Construct a timeline segment on the scheduling side and compare it one-to-one with the vector segment generated by the main server; S6-2. In the comparison results, vector segments with an overlap greater than 95% and a deviation less than 1% are marked as valid. S6-3. The remaining vector segments are marked as segments to be backtracked and handed over to the asynchronous data consistency engine for fine-tuning and reconstruction.

[0012] Preferably, the edge node topology map construction of S7 includes obtaining the geographical coordinates, communication link quality and signal delay information of all power plant edge collection points, establishing a directed topology matrix, determining the influence factor of each node on the abnormal segment according to the signal propagation order and power contribution relationship, and embedding the topology factor into the power vector structure for multi-dimensional correlation analysis, so as to realize the positioning isolation of the abnormal area and generate statistical repair suggestions.

[0013] Preferably, the edge node topology map construction in S7 further includes the following steps: S7-1 obtains the link state diagram between the scheduling synchronization nodes, and dynamically assigns weights to each edge, and the weight values ​​are updated in real time based on the link quality; S7-2. Based on the power flow tracking algorithm, reverse the abnormal propagation path of the abnormal power segment and build a path tree; S7-3. Set a multi-hop isolation threshold on the path tree. When the propagation path length exceeds 3 hops, execute chain isolation marking and output the traceability results for the path-related nodes in the statistical report.

[0014] Compared with the prior art, the advantages of the present invention are: (1) The introduction of PPS physical triggering + temperature-controlled crystal oscillator compensation + multi-source atomic clock backup significantly improves the stability of timestamps and avoids the millisecond-level drift common in SCADA systems.

[0015] (2) Innovatively adopt 5G and Ethernet parallel transmission paths to improve the anti-interference and disaster recovery capabilities of power collection, which is different from the traditional single-link transmission architecture.

[0016] (3) Quintic spline interpolation and 128-dimensional electric quantity vector expression are used to achieve high-density electric quantity feature compression, improve statistical efficiency and traceability, and distinguish it from the original sequence statistics.

[0017] (4) Using SHA-512 to construct a forward and reverse chain structure to identify non-continuous segments, it has stronger integrity verification capabilities compared to the existing method of using only timestamp comparison.

[0018] (5) By constructing a directed topology matrix and combining it with node power weights, a multi-dimensional mapping of abnormal segments and edge nodes is achieved, breaking through the traditional point-to-point traceability method.

[0019] (6) A joint standard of vector dimension coincidence and error rate is adopted to replace the traditional method of comparing the absolute value of electric quantity, thereby improving the accuracy and stability of cross-system time alignment.

[0020] (7) Set the propagation path hop count threshold to achieve link-level isolation and path freezing of abnormal segments, thereby enhancing system security and grid abnormality response capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is a schematic diagram of the overall system of a power plant electricity statistics method according to the present invention; DETAILED DESCRIPTION

[0022] For examples, see Figure 1 A power plant electricity statistics method includes the following steps: S1. Based on the intelligent edge gateway device on the generator side, the analog signals of current, voltage, frequency and power factor are collected and the first timestamp correction is performed; S2. Upload the corrected signal simultaneously to the power master server and the scheduling synchronization node via a dual-channel path; S3. Build minute-level electricity data blocks on the main electricity server, reconstruct the boundary errors of adjacent time periods by taking the difference, and generate a unified format of time-series electricity vectors. S4. Perform bidirectional hash tracking on each time series power vector to filter out non-continuous power segments; S5. Perform structured analysis on the time series electricity, extract the power generation load change segments, and generate a partitioned statistical report; S6. Based on the global power timeline generated by the scheduling synchronization node, the statistical results of the power master server are synchronized and verified to build a reliable daily power statistical report; S7. Construct an edge node topology map and associate the constructed edge node topology map with statistical data for topological positioning and isolation control of abnormal power segments.

[0023] Specifically, a power plant electricity statistics system includes a generator-side intelligent edge gateway, an electricity master server, a scheduling synchronization node, and a data management platform, wherein: The generator-side edge gateway uses a TI-AM64x industrial edge device with a 32-bit analog-to-digital converter (ADC) module, a 16-bit sampling accuracy, and a 20kHz sampling frequency. The main power server is a XeonSilver multi-core processor host, deployed with a time series database (such as InfluxDB) and a parallel processing engine; The scheduling synchronization node is equipped with an independent atomic clock module and a GPS receiver with a PPS interface (such as U-BloxZED-F9T); All modules are connected via industrial Ethernet and independent 5G links, supporting the MQTT protocol and link intermittent retransmission function.

[0024] Specifically, after collecting parameters such as current, voltage, frequency, and power factor, the edge gateway converts them into digital signals via ADC, performs timestamp correction based on sliding mean and temperature compensation (see claims 2-3 for details), and then uploads them to the power master server and scheduling synchronization node via dual-path transmission. The power master server constructs one power data block per minute, constructs vector segments through quintic spline interpolation (see claims 4-5 for details), and forms a 128-dimensional time series power vector in a unified format. A global timeline is formed on the scheduling node side for cross-validation of the master server results. The system supports anomaly identification and power repair based on topology maps.

[0025] The first timestamp correction process of S1 further includes the following steps: S1-1. Align the system clock of the intelligent edge gateway device on the generator side with the external timing module through the PPS physical trigger signal, and combine it with the temperature-controlled crystal oscillator to stabilize the delay response; S1-2. Apply the sliding time window mean correction method to fine-tune and compensate the local time offset of the continuously sampled signal to eliminate the timing error caused by dynamic drift.

[0026] This implementation uses the GPS timing module U-BloxZED-F9T for physical-level PPS alignment with the edge device's built-in TCXO (temperature-controlled crystal oscillator). A sliding average algorithm is used to filter the time offset of consecutive samples in a three-point window, with a drift limit of ±3 μs. If the drift threshold is exceeded, a dynamic fine-tuning mechanism is automatically implemented.

[0027] S1-1 also includes the following steps: S1-1-1. Introducing a multi-source redundant timing mechanism that automatically switches to a backup atomic clock signal when the primary clock signal is interrupted. S1-1-2. Set the hard trigger threshold window. The response delay of less than 2 microseconds is directly marked as the high-precision authorization period. S1-1-3. All timing segments are time-traced and labeled to ensure that subsequent abnormal periods can be traced back to the corresponding hardware offset node.

[0028] Specifically, if the master clock is lost, the edge gateway automatically switches to a rubidium atomic clock or a backup satellite timing system. Simultaneously, a 2μs threshold check is performed on all PPS trigger periods, marking stable segments as high-precision segments. All timestamp chains record their timing source, drift rate, and correction value, forming a traceable index field for subsequent anomaly location.

[0029] Generating S3 minute-level power data blocks also includes the following steps: S3-1. Normalize the signal data within each 60-second segment and accumulate the raw data segments of equal length; S3-2. Use quintic spline interpolation to generate a smooth transition curve for each edge point, and perform continuity verification based on the curvature of the center segment; S3-3. Encode each piece of processed data into a 128-dimensional vector, retaining the original node index information as a verification field for subsequent chain tracing.

[0030] Specifically, the raw data, which contains 1.2 million sampling points per minute, is divided into 60 segments, each with 20,000 points. These segments are normalized and then merged into the original segments. The segments are interpolated using quintic spline interpolation with a maximum curvature of 2.0. Each segment is compressed into a 128-dimensional vector, and the node number, timestamp, and fitting residual are recorded.

[0031] The interpolation reconstruction of S3-2 also includes the following steps: S3-2-1. Select 15 sampling points before and after the boundary within the time window as fitting reference points; S3-2-2. Introduce a curvature control factor to limit the rate of change of the derivative after fitting to be less than the set gradient threshold; S3-2-3. Make secondary adjustments to the interpolation segment based on the actual frequency offset to eliminate the fitting error caused by grid frequency disturbance.

[0032] Specifically, the boundary point fitting is based on the 15 points before and after, and the interpolate.splrep function in Python Scipy is used to generate the fitting function. The curvature control factor is λ = 0.003. If the residual is greater than 0.01, the frequency perturbation secondary adjustment is performed. The error after reconstruction is controlled within ±0.5%.

[0033] S4's bidirectional hash tracking process also includes the following steps: S4-1. Execute the SHA-512 algorithm to encode the time series power vector to form a forward linked list structure; S4-2. Encode the same data in reverse order into a reverse chain and perform hash value verification on each node; S4-3. If the chain continuity is interrupted, mark the interruption point index and isolate the two discontinuous segments before and after.

[0034] Specifically, each charge vector is encoded using SHA-512 to generate a forward chain, and its reverse data is also encoded using the same SHA-512 encoding to form a reverse chain. The two chains are cross-compared, and hash consistency is verified within each time window. If the hash value of a breakpoint node does not match, its index and the previous and next fragments are recorded and pushed to the traceability module for isolation and repair.

[0035] The synchronization verification process of S6 also includes the following steps: S6-1. Construct a timeline segment on the scheduling side and compare it one by one with the vector segment generated by the power master server; S6-2. In the comparison results, vector segments with an overlap greater than 95% and a deviation less than 1% are marked as valid. S6-3. The remaining vector segments are marked as segments to be backtracked and handed over to the asynchronous data consistency engine for fine-tuning and reconstruction.

[0036] Specifically, the scheduling node matches the global timeline segments generated every 5 minutes with the power vector segments generated by the master server segment by segment. The matching method is dimension-by-dimensional error comparison (threshold 1%). Segments with a matching degree > 95% are considered valid; the remaining segments are marked as to be reconstructed and sent to the consistency engine via Kafka streaming to perform historical segment compensation.

[0037] The construction of S7's edge node topology map involves obtaining the geographic coordinates, communication link quality, and signal delay information of all power plant edge collection points, establishing a directed topology matrix, determining the impact factor of each node on the abnormal segment according to the signal propagation order and power contribution relationship, and embedding the topology factor into the power vector structure for multi-dimensional correlation analysis, thereby achieving the positioning and isolation of abnormal areas and generating statistical repair suggestions.

[0038] Specifically, all edge gateways record their latitude and longitude locations and uplink node addresses during installation. The system then constructs an edge-node relationship matrix based on link stability (packet loss rate < 0.1%) and average latency. The power impact of each node is calculated using the following formula: W_i=(P_i / ΣP)*(1 / D_i) Where: W_i is the topology factor of node i, P_i is the average power of node i, and D_i is the average propagation delay to the master node.

[0039] The construction of the edge node topology map in S7 also includes the following steps: S7-1. Obtain the link state diagram between the scheduling synchronization nodes and dynamically assign weights to each edge. The weight values ​​are updated in real time based on the link quality. S7-2. Based on the power flow tracking algorithm, reverse the abnormal propagation path of the abnormal power segment and build a path tree; S7-3. Set a multi-hop isolation threshold on the path tree. When the propagation path length exceeds 3 hops, execute chain isolation marking and output the traceability results for the path-related nodes in the statistical report.

[0040] Specifically, the topological link state graph is represented using the GraphML structure, and the weight of each edge is updated according to the measured link packet loss rate; In power flow tracing, the BFS algorithm is used to traverse backward to the abnormal node and construct a path tree; The multi-hop isolation threshold is set at 3 hops. If the path length is exceeded, the path freeze is automatically triggered, and the nodes involved in the path are marked in red in the statistical report, allowing power supply network operation and maintenance personnel to predict risk sources.

[0041] The above shows and describes the basic principles, main features and advantages of the present invention; those skilled in the art should understand that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements all fall within the scope of the present invention to be protected; the scope of protection claimed in the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for statistically analyzing power consumption in a power plant, characterized in that: The power plant electricity statistics method comprises the following steps: S1. Based on the intelligent edge gateway device on the generator side, the analog signals of current, voltage, frequency and power factor are collected and the first timestamp correction is performed; S2. Upload the corrected signal simultaneously to the power master server and the scheduling synchronization node via a dual-channel path; S3. Construct minute-level electricity data blocks on the main power server side, reconstruct the boundary errors of adjacent time periods by difference, and generate a time series electricity vector in a unified format; S4. Perform bidirectional hash tracking processing on each of the time series power vectors to filter out non-continuous power segments; S5. Perform structured analysis on the time series electricity, extract the power generation load change section, and generate a partition statistical report; S6. Based on the global power timeline generated by the scheduling synchronization node, the statistical results of the power master server are synchronized and verified to build a reliable daily power statistics report; S7. Construct an edge node topology map and associate the constructed edge node topology map with statistical data for topological positioning and isolation control of abnormal power segments.

2. A power plant electricity statistics method according to claim 1, characterized in that: The first timestamp correction process of S1 further includes the following steps: S1-1. The system clock of the generator side smart edge gateway device is aligned with the external timing module through the PPS physical trigger signal, and combined with a temperature-controlled crystal oscillator to stabilize the delayed response; S1-2. Apply the sliding time window mean correction method to fine-tune and compensate the local time offset of the continuously sampled signal to eliminate the timing error caused by dynamic drift.

3. A power plant electricity statistics method according to claim 2, characterized in that: The S1-1 further comprises the following steps: S1-1-1. Introducing a multi-source redundant timing mechanism that automatically switches to a backup atomic clock signal when the primary clock signal is interrupted. S1-1-2. Set the hard trigger threshold window. The response delay of less than 2 microseconds is directly marked as the high-precision authorization period. S1-1-3. All timing segments are time-traced and labeled to ensure that subsequent abnormal periods can be traced back to the corresponding hardware offset node.

4. A power plant electricity statistics method according to claim 1, characterized in that: The generation of the minute-level electricity data block of S3 further includes the following steps: S3-1. Normalize the signal data within each 60-second segment and accumulate the raw data segments of equal length; S3-2. Use quintic spline interpolation to generate a smooth transition curve for each edge point, and perform continuity verification based on the curvature of the center segment; S3-3. Encode each piece of processed data into a 128-dimensional vector, retaining the original node index information as a verification field for subsequent chain tracing.

5. A power plant electricity statistics method according to claim 4, characterized in that: The interpolation reconstruction of S3-2 further includes the following steps: S3-2-1. Select 15 sampling points before and after the boundary within the time window as fitting reference points; S3-2-2. Introduce a curvature control factor to limit the rate of change of the derivative after fitting to be less than the set gradient threshold; S3-2-3. Make secondary adjustments to the interpolation segment based on the actual frequency offset to eliminate the fitting error caused by grid frequency disturbance.

6. A power plant electricity statistics method according to claim 1, characterized in that: The bidirectional hash tracking process of S4 further includes the following steps: S4-1. Execute SHA-512 algorithm encoding to form a forward linked list structure on the timing power vector; S4-2. Encode the same data in reverse order into a reverse chain and perform hash value verification on each node; S4-3. If the chain continuity is interrupted, mark the interruption point index and isolate the two discontinuous segments before and after.

7. A power plant electricity statistics method according to claim 1, characterized in that: The synchronization verification process of S6 further includes the following steps: S6-1. Construct a timeline segment on the scheduling side and compare it one-to-one with the vector segment generated by the main server; S6-2. In the comparison results, vector segments with an overlap greater than 95% and a deviation less than 1% are marked as valid. S6-3. The remaining vector segments are marked as segments to be backtracked and handed over to the asynchronous data consistency engine for fine-tuning and reconstruction.

8. The power plant electricity statistics method according to claim 1, characterized in that: The construction of the S7 edge node topology map includes obtaining the geographic coordinates, communication link quality and signal delay information of all power plant edge collection points, establishing a directed topology matrix, determining the influence factor of each node on the abnormal segment according to the signal propagation order and power contribution relationship, and embedding the topology factor into the power vector structure for multi-dimensional correlation analysis, so as to achieve the positioning isolation of the abnormal area and generate statistical repair suggestions.

9. A power plant electricity statistics method according to claim 8, characterized in that: The edge node topology map construction in S7 further includes the following steps: S7-1 obtains the link state diagram between the scheduling synchronization nodes, and dynamically assigns weights to each edge, and the weight values ​​are updated in real time based on the link quality; S7-2. Based on the power flow tracking algorithm, reverse the abnormal propagation path of the abnormal power segment and build a path tree; S7-3. Set a multi-hop isolation threshold on the path tree. When the propagation path length exceeds 3 hops, execute chain isolation marking and output the traceability results for the path-related nodes in the statistical report.

Images