Hydropower station operation data intelligent management system and method based on lake and warehouse integration

Abstract

The application discloses a hydropower station operation data intelligent management system and method based on lake-warehouse integration, relates to the technical field of hydropower station management, and collects multi-source operation data of the hydropower station, carries out data cleaning, abnormality detection and normalization preprocessing, generates standardized time series data, and stores the standardized time series data in a lake-warehouse integrated platform; based on the preprocessed data, the decision value characteristics of each data point are analyzed, and the value index is analyzed based on the decision value characteristic mode; according to the value index, the calculation and storage resources of the system are dynamically allocated; when it is identified that the operation state deviates from a preset optimal interval, root cause analysis is carried out based on high-value data, and after the decision data authorization execution, the hydropower station operation state is guided to return to the optimal interval according to the deviation reduction rate. The real-time response capability of the system is improved; and the stability and robustness of the system under high load are improved.

Description

Technical Field

[0001] This invention relates to the field of hydropower station management technology, specifically to an intelligent management system and method for hydropower station operation data based on the integration of lake and reservoir. Background Technology

[0002] Existing hydropower station operation optimization and fault handling systems can provide basic over-limit alarms, but managers often need to spend a significant amount of time manually sifting through data and troubleshooting fault points. This process is slow and susceptible to subjective factors. Furthermore, computing and storage resources are typically statically configured, meaning that when resources are scarce, the system may not be able to prioritize critical analysis tasks, while processing large amounts of low-value data results in unnecessary resource waste. This lack of dynamic resource allocation and intelligent closed-loop control makes it difficult for hydropower stations to continuously maintain their optimal operating range, hindering further improvements in their economic efficiency and safety levels.

[0003] Therefore, this invention discloses an intelligent management system and method for hydropower station operation data based on the integration of lake and reservoir to solve the above problems. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent management system and method for hydropower station operation data based on lake-warehouse integration, so as to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an intelligent management method for hydropower station operation data based on integrated lake and reservoir system, the method comprising the following steps:

[0006] S1: Collect multi-source operation data of hydropower stations, perform data cleaning, anomaly detection and normalization preprocessing, generate standardized time series data, and store them in the integrated lake and warehouse platform;

[0007] S2: Based on the preprocessed data, analyze the decision value characteristics of each data point, and analyze the value indicators based on the decision value characteristics.

[0008] S3: Dynamically allocate the system's computing and storage resources based on value indicators;

[0009] S4: When the operating status is found to deviate from the preset optimal range, root cause analysis is performed based on high-value data. After the decision data is authorized for execution, the operating status of the hydropower station is guided back to the optimal range according to the deviation reduction rate.

[0010] According to the above scheme, S1 includes the following:

[0011] S101: Real-time acquisition of hydropower station operation data through sensor network. The hydropower station operation data includes water level data, flow rate data, equipment power data, equipment temperature data and equipment vibration data, forming a raw data stream, wherein each data point contains timestamp and spatial location information.

[0012] S102: Clean the raw data stream by using a Z-score-based anomaly detection method to identify and remove outliers; normalize the various types of data in the data stream after outlier removal to generate standardized data; store the standardized data in the lake-warehouse integrated platform, including a data lake for raw storage and a data warehouse for structured queries, and output standardized time-series data.

[0013] This invention collects multi-dimensional data, including water level, flow rate, power, temperature, and vibration, through a sensor network, along with timestamps and spatial location information. This provides a complete and traceable data foundation for subsequent refined analysis and spatial correlation calculations. It adopts a hybrid architecture combining a data lake and a data warehouse, which can both use the data lake to store raw data at low cost for future in-depth mining, and use the efficient query capabilities of the data warehouse to serve real-time analysis, achieving the best balance between cost and performance.

[0014] According to the above scheme, S2 includes the following:

[0015] S201: Obtain the set of spatial neighbor nodes for each monitoring node in the standardized time series data. Denote the set of spatial neighbor nodes for the i-th monitoring node as N(i). Analyze the feature similarity sim(d) between monitoring node i and its spatial neighbor nodes. i d j ); where d i d represents the running data vector of monitoring node i. j sim(d) represents the running data vector of the j-th monitoring node in the set of spatial neighbor nodes. i d j =exp(-||d) i -d j || 2 / (2σ) 2 )); where ||d i -d j || represents the running data vector d i and running data vector d j The Euclidean distance between them; σ represents the preset bandwidth parameter; exp() represents the exponential function with the natural number base; the spatial correlation S of monitoring node i is analyzed based on the feature similarity between monitoring node i and each of its spatial neighbors. i ,

[0016] ;

[0017] Where N represents the total number of monitoring nodes in the spatial neighbor node set N(i); w （i，j） w represents the spatial association weight between monitoring node i and monitoring node j. （i，j） =1 / (1+d （i，j） ); where d （i，j） This represents the Euclidean distance between monitoring node i and monitoring node j;

[0018] S202: Analyze data freshness based on the difference between the current time and the data generation time; denot the data freshness of monitoring node i as TN. i TN i =exp(-λ△t i ); where λ represents the preset attenuation factor; Δt i It represents the difference between the current time and the data generation time; the ratio between the change in value and the maximum change in value is recorded as the data change rate; and a weighted sum of the data freshness and data urgency is generated to produce a time urgency index.

[0019] S203: Analyze the decision contribution based on the similarity between the monitoring point's operational data vector and the historical decision-related operational data vector, and denote the decision contribution of monitoring node i as C. i ,

[0020] ;

[0021] Where M represents the number of monitoring nodes corresponding to historical decisions, d m Let U represent the operational data vector of the monitoring node corresponding to the m-th historical decision, f() represent the cosine similarity function, and U m The quantitative value representing the degree of state improvement corresponding to the m-th historical decision is, in this embodiment, the quantitative value of the degree of state improvement is the power deviation reduction rate.

[0022] S204: Combining spatial relevance, time urgency, and decision contribution, a weighted summation method is used to analyze value indicators.

[0023] This invention quantifies the consistency between a monitoring point and its neighbors by calculating spatial correlation, thereby quickly locating local anomalies or identifying the scope of fault propagation. By using data freshness and time urgency indicators, the system can prioritize processing the latest and most drastically changing data, ensuring that decisions are based on the latest situation and improving the system's real-time response capability. By assigning value weights based on the contribution of historical data to improving the operational status, the value assessment is not only based on the current state but also incorporates historical experience, improving the accuracy of the assessment.

[0024] According to the above scheme, S3 includes the following:

[0025] S301: Based on the statistical distribution of historical value indicators, a dynamic classification threshold is set using the percentile method; data with value indicators greater than or equal to the high value threshold are classified as high value group, data with value indicators less than or equal to the low value threshold are classified as low value group, and data with value indicators less than the high value threshold but greater than the low value threshold are classified as medium value group.

[0026] S302: For high-value groups, allocate a memory database at the storage level to ensure millisecond-level read and write access to the data; at the computation level, call a complex algorithm for processing; the complex algorithm includes a physical mechanism simulation model based on the turbine characteristic curve, generator efficiency spectrum and fluid dynamics equations, and an unscented Kalman filter;

[0027] For low-value groups, an arithmetic mean aggregation method is used to average every certain number of data points, and downsampling is implemented for storage; at the computational level, a lightweight algorithm is used; the lightweight algorithm includes moving average and exponential smoothing;

[0028] S303: Monitor system resource utilization rate, and adjust the high-value threshold based on the target resource utilization rate and the real-time resource utilization rate. The resource utilization rate can be the CPU or memory utilization rate.

[0029] V high，new =V high，old +k a (U target -U)+k b ∫(U target -U)dt; where V high，new This indicates the adjustment of the high-value threshold, V high，old Indicates the original high-value threshold; k a and k b U represents the control gain coefficient; target denoted as target resource utilization rate, and U represents real-time resource utilization rate; the low value threshold is corrected based on the high value threshold; the difference between the corrected low value threshold and the original low value threshold is equal to the difference between the corrected high value threshold and the original high value threshold.

[0030] This invention, based on precise value indicators, implements differentiated processing strategies for high- and low-value data, ensuring that limited system resources are prioritized for the most critical data in decision-making, thereby achieving overall cost reduction and efficiency improvement. High-precision complex models (such as physical simulation and unscented Kalman filtering) are used for high-value data to guarantee the accuracy of core decisions; lightweight models and downsampling are used for low-value data, effectively saving resources. Through a feedback adjustment mechanism, the system can dynamically adjust the value threshold based on real-time resource utilization, ensuring the system's stability and robustness under high load.

[0031] According to the above scheme, S4 includes the following:

[0032] S401: Continuously monitor the operating status indicators. If the deviation of the operating status indicator exceeds the preset tolerance threshold, it is determined that the operating status has deviated from the optimal range, and a status deviation signal is output. The status deviation signal includes a deviation Boolean value, parameter identifier, deviation amount, and timestamp. Extract all high-value data corresponding to the timestamp of the detected deviation. Analyze the mutual information between each potential influencing factor and the deviation signal based on the high-value data. Mark the potential influencing factor with the largest mutual information value as the root cause of the current operating status deviation and generate an early warning signal.

[0033] S402: Extract decision data and send it to the administrator to request authorization. After the decision data is authorized and executed, compare the actual state with the expected state to calculate the reduction rate of deviation of the operating status index. The reduction rate of deviation of the operating status index is equal to the ratio of the difference between the deviation of the operating status index before and after the adjustment to the deviation of the operating status index before the adjustment. If the deviation reduction rate is less than the minimum improvement threshold, trigger the re-analysis process, use the latest data to re-perform root cause analysis and generate a new decision flow.

[0034] This invention employs a mutual information method based on information theory for root cause analysis, which can capture nonlinear relationships between variables. It is more suitable for complex hydropower systems than traditional linear correlation analysis methods, and its ability to pinpoint root causes is more accurate. Decision data requires administrator authorization for execution, ensuring that human experts have ultimate control over key decisions and improving the system's security and reliability. The post-execution deviation reduction rate verification and re-analysis mechanism forms a closed loop of "planning-execution-checking-action," enabling the system to learn from errors and continuously self-optimize, ultimately achieving autonomous and reliable return to the optimal operating range.

[0035] Another aspect of this application provides an intelligent management system for hydropower station operation data based on the integration of lake and reservoir. The system is applied to the above-mentioned intelligent management method for hydropower station operation data based on the integration of lake and reservoir. The system includes a hydropower station data processing module, a hydropower station feature analysis module, a resource allocation module, and an operation monitoring and optimization module.

[0036] The hydropower station data processing module is used to collect multi-source operation data of the hydropower station, and perform data cleaning, anomaly detection and normalization preprocessing to generate standardized time series data, which is stored in the lake-warehouse integrated platform.

[0037] The hydropower station feature analysis module is used to analyze the decision value characteristics of each data point based on the preprocessed data, and to analyze value indicators based on the decision value characteristics.

[0038] The resource allocation module is used to dynamically allocate the system's computing and storage resources based on value indicators;

[0039] The operation monitoring and optimization module is used to perform root cause analysis based on high-value data when the operation status is detected to deviate from the preset optimal range. After the decision data is authorized to be executed, the module guides the hydropower station's operation status back to the optimal range based on the deviation reduction rate.

[0040] According to the above scheme, the hydropower station data processing module includes a hydropower station data acquisition unit and a hydropower station data integration unit;

[0041] The hydropower station data acquisition unit is used to collect the hydropower station's operating data in real time through a sensor network. The hydropower station's operating data includes water level data, flow rate data, equipment power data, equipment temperature data, and equipment vibration data, forming a raw data stream.

[0042] The hydropower station data integration unit is used to clean the raw data stream, identify and remove outliers using a Z-score-based anomaly detection method, normalize various types of data in the data stream after outlier removal to generate standardized data, store the standardized data in the lake-warehouse integrated platform, and output standardized time series data.

[0043] According to the above scheme, the hydropower station feature analysis module includes a sub-feature analysis unit and a feature fusion unit;

[0044] The sub-feature analysis unit is used to monitor the spatial correlation, time urgency index, and decision contribution of nodes.

[0045] The feature fusion unit combines spatial correlation, time urgency, and decision contribution indicators, and uses a weighted summation method to analyze value indicators.

[0046] According to the above scheme, the resource allocation module includes a data partitioning unit, an allocation unit, and a threshold correction unit;

[0047] The data segmentation unit is used to set dynamic classification thresholds based on the statistical distribution of historical value indicators and using the percentile method; data with value indicators greater than or equal to the high value threshold are classified into the high value group, data with value indicators less than or equal to the low value threshold are classified into the low value group, and data with value indicators less than the high value threshold but greater than the low value threshold are classified into the medium value group.

[0048] The allocation unit allocates a memory database at the storage level for high-value groups to ensure millisecond-level read and write access to the data; at the computation level, it calls a complex algorithm for processing; the complex algorithm includes a physical mechanism simulation model based on turbine characteristic curves, generator efficiency spectra, and fluid dynamics equations, and an unscented Kalman filter; for low-value groups, it uses an arithmetic mean aggregation method to average every certain number of data points and implements downsampling storage; at the computation level, it employs a lightweight algorithm; the lightweight algorithm includes moving average and exponential smoothing;

[0049] The threshold correction unit is used to monitor the system resource utilization rate, correct the high-value threshold based on the target resource utilization rate and the real-time resource utilization rate, and correct the low-value threshold based on the corrected high-value threshold; the difference between the corrected low-value threshold and the original low-value threshold is equal to the difference between the corrected high-value threshold and the original high-value threshold.

[0050] According to the above scheme, the operation monitoring and optimization module includes a status analysis unit and a deviation analysis unit;

[0051] The status analysis unit is used to continuously monitor the operating status indicators. If the deviation of the operating status indicator exceeds the preset tolerance threshold, it determines that the operating status has deviated from the optimal range and outputs a status deviation signal. It extracts all high-value data corresponding to the timestamp of the detected deviation. Based on the high-value data, it analyzes the mutual information between each potential influencing factor and the deviation signal. It marks the potential influencing factor with the largest mutual information value as the root cause of the current operating status deviation and generates an early warning signal.

[0052] The deviation analysis unit is used to extract decision data and send it to the administrator to request authorization. After the decision data is authorized and executed, the actual state is compared with the expected state to calculate the deviation reduction rate. If the deviation reduction rate is less than the minimum improvement threshold, the re-analysis process is triggered to re-analyze the root cause using the latest data and generate a new decision flow.

[0053] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention collects multi-dimensional data including water level, flow rate, power, temperature, and vibration through a sensor network, along with timestamps and spatial location information, providing a complete and traceable data foundation for subsequent refined analysis and spatial correlation calculations. It adopts a hybrid architecture combining a data lake and a data warehouse, which can utilize the data lake to store raw data at low cost for future in-depth analysis, while leveraging the efficient query capabilities of the data warehouse to serve real-time analysis, achieving an optimal balance between cost and performance. By calculating spatial correlation, this invention can quantify the consistency between the state of a monitoring point and its neighbors, thereby quickly locating local anomalies or identifying the propagation range of faults. Through data freshness and time urgency indicators, the system can prioritize processing the latest and most drastically changing data, ensuring that decisions are based on the latest situation and improving the system's real-time response capability. Through decision contribution, it assigns value weights based on the contribution of historical data to improving operational status, making value assessment not only based on the current state but also incorporating historical experience, thus improving the accuracy of the assessment. This invention, based on precise value indicators, implements differentiated processing strategies for high- and low-value data, ensuring that limited system resources are prioritized for the most critical data in decision-making, thereby achieving overall cost reduction and efficiency improvement. High-value data employs high-precision complex models to guarantee the accuracy of core decisions; low-value data utilizes lightweight models and downsampling, effectively conserving resources. Through a feedback adjustment mechanism, the system can dynamically adjust value thresholds based on real-time resource utilization, ensuring system stability and robustness under high loads. This invention uses a mutual information method based on information theory for root cause analysis, capable of capturing nonlinear relationships between variables, making it more suitable for complex hydropower systems than traditional linear correlation analysis methods, and more accurately locating root causes. Decision data requires administrator authorization for execution, ensuring human experts have ultimate control over key decisions, improving system security and reliability. The post-execution deviation reduction rate verification and re-analysis mechanism forms a closed loop of "planning-execution-checking-action," enabling the system to learn from errors and continuously self-optimize, ultimately achieving autonomous and reliable return to the optimal operating range. Attached Figure Description

[0054] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0055] Figure 1 This is a flowchart illustrating the intelligent management method for hydropower station operation data based on the integration of lake and reservoir systems, as described in this invention.

[0056] Figure 2 This is a schematic diagram of the intelligent management system for hydropower station operation data based on the integration of lake and reservoir, as proposed in this invention. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] Please see Figure 1 This invention provides a technical solution: an intelligent management method for hydropower station operation data based on integrated lake and reservoir system, which includes the following steps:

[0059] S1: Collect multi-source operation data of hydropower stations, perform data cleaning, anomaly detection and normalization preprocessing, generate standardized time series data, and store them in the integrated lake and warehouse platform;

[0060] S1 includes the following:

[0061] S101: Real-time acquisition of hydropower station operation data through sensor network. The hydropower station operation data includes water level data, flow rate data, equipment power data, equipment temperature data and equipment vibration data, forming a raw data stream, in which each data point contains timestamp and spatial location information.

[0062] S102: Clean the raw data stream by using a Z-score-based anomaly detection method to identify and remove outliers; normalize the various types of data in the data stream after outlier removal to generate standardized data; store the standardized data in the lake-warehouse integrated platform, including a data lake for raw storage and a data warehouse for structured queries, and output standardized time-series data.

[0063] S2: Based on the preprocessed data, analyze the decision value characteristics of each data point, and analyze the value indicators based on the decision value characteristics.

[0064] S2 includes the following:

[0065] S201: Obtain the set of spatial neighbor nodes for each monitoring node in the standardized time series data. Denote the set of spatial neighbor nodes for the i-th monitoring node as N(i). Analyze the feature similarity sim(d) between monitoring node i and its spatial neighbor nodes. i d j ); where d i d represents the running data vector of monitoring node i. j sim(d) represents the running data vector of the j-th monitoring node in the set of spatial neighbor nodes. i d j=exp(-||d) i -d j || 2 / (2σ) 2 )); where ||d i -d j || represents the running data vector d i and running data vector d j The Euclidean distance between them; σ represents the preset bandwidth parameter; exp() represents the exponential function with the natural number base; the spatial correlation S of monitoring node i is analyzed based on the feature similarity between monitoring node i and each of its spatial neighbors. i ,

[0066] ;

[0067] Where N represents the total number of monitoring nodes in the spatial neighbor node set N(i); w （i，j） w represents the spatial association weight between monitoring node i and monitoring node j. （i，j） =1 / (1+d （i，j） ); where d （i，j） This represents the Euclidean distance between monitoring node i and monitoring node j;

[0068] S202: Analyze data freshness based on the difference between the current time and the data generation time; denot the data freshness of monitoring node i as TN. i TN i =exp(-λ△t i ); where λ represents the preset attenuation factor; Δt i It represents the difference between the current time and the data generation time; the ratio between the change in value and the maximum change in value is recorded as the data change rate; and a weighted sum of the data freshness and data urgency is generated to produce a time urgency index.

[0069] S203: Analyze the decision contribution based on the similarity between the monitoring point's operational data vector and the historical decision-related operational data vector, and denote the decision contribution of monitoring node i as C. i ,

[0070] ;

[0071] Where M represents the number of monitoring nodes corresponding to historical decisions, d m Let U represent the operational data vector of the monitoring node corresponding to the m-th historical decision, f() represent the cosine similarity function, and U m The quantitative value representing the degree of state improvement corresponding to the m-th historical decision is, in this embodiment, the quantitative value of the degree of state improvement is the power deviation reduction rate.

[0072] S204: Combining spatial relevance, time urgency, and decision contribution, a weighted summation method is used to analyze value indicators.

[0073] S3: Dynamically allocate the system's computing and storage resources based on value indicators;

[0074] S3 includes the following:

[0075] S301: Based on the statistical distribution of historical value indicators, a dynamic classification threshold is set using the percentile method; data with value indicators greater than or equal to the high value threshold are classified as high value group, data with value indicators less than or equal to the low value threshold are classified as low value group, and data with value indicators less than the high value threshold but greater than the low value threshold are classified as medium value group.

[0076] S302: For high-value groups, allocate an in-memory database at the storage level to ensure millisecond-level read and write access to the data; at the computation level, call complex algorithms for processing; the complex algorithm type includes a physical mechanism simulation model based on turbine characteristic curves, generator efficiency spectra and fluid dynamics equations, and an unscented Kalman filter;

[0077] For low-value groups, an arithmetic mean aggregation method is used to average every certain number of data points, and downsampling is implemented for storage; at the computational level, lightweight algorithms are used; lightweight algorithms include moving average and exponential smoothing;

[0078] S303: Monitor system resource utilization rate, adjust high-value thresholds based on target resource utilization rate and real-time resource utilization rate, V high，new =V high，old +k a (U target -U)+k b ∫(U target -U)dt; where V high，new This indicates the adjustment of the high-value threshold, V high，old Indicates the original high-value threshold; k a and k b U represents the control gain coefficient; target denoted as target resource utilization rate, and U represents real-time resource utilization rate; the low value threshold is corrected based on the high value threshold; the difference between the corrected low value threshold and the original low value threshold is equal to the difference between the corrected high value threshold and the original high value threshold.

[0079] S4: When the operating status is found to deviate from the preset optimal range, root cause analysis is performed based on high-value data. After the decision data is authorized for execution, the operating status of the hydropower station is guided back to the optimal range according to the deviation reduction rate.

[0080] Example 1: In S4, the following is included:

[0081] S401: Continuously monitors operating status indicators. If the deviation of the operating status indicator exceeds the preset tolerance threshold, it is determined that the operating status has deviated from the optimal range and outputs a status deviation signal.

[0082] Operating status index deviations include, but are not limited to, power deviation, water level deviation, and efficiency deviation. In this embodiment, power deviation is taken as an example of operating status index deviation. In this embodiment, the optimal power range of Unit 1 of the hydropower station is [95, 105] MW. Then, the midpoint value P of the optimal power range is... optimal =100MW,

[0083] At time T1: The system monitors the actual power P actual =102MW; Calculate the operating status index deviation, i.e., the power deviation, as ΔP=|102−100|=2MW; Tolerance threshold=100-95=5MW; If 2<5, do not trigger the status deviation signal, and the system remains in monitoring state;

[0084] At time T2: The system monitors the actual power P actual =88MW; calculate the operating status index deviation, i.e., the power deviation, as ΔP=|88−100|=12MW; determine that 12>5, and trigger the generation of the status deviation signal;

[0085] The status deviation signal includes a deviation Boolean value, parameter identifier, deviation amount, and timestamp; extract all high-value data corresponding to the timestamp of the detected deviation; analyze the mutual information between each potential influencing factor and the deviation signal based on the high-value data; mark the potential influencing factor with the largest mutual information value as the root cause of the current operating status deviation and generate an early warning signal;

[0086] S402: Extract decision data and send it to the administrator to request authorization. After the decision data is authorized and executed, compare the actual state with the expected state to calculate the reduction rate of deviation of the operating status index. The reduction rate of deviation of the operating status index is equal to the ratio of the difference between the deviation of the operating status index before and after the adjustment to the deviation of the operating status index before the adjustment.

[0087] In this embodiment, power is used as the operating status indicator; therefore, the deviation of the operating status indicator is the power deviation, and the reduction rate of the operating status indicator deviation is the power deviation reduction rate.

[0088] Power deviation reduction rate η = (ΔP) before -ΔP after ) / ΔP before ;where ΔP before ΔP represents the power deviation before adjustment. after This indicates the adjusted power deviation;

[0089] If the deviation reduction rate is less than the minimum improvement threshold, the re-analysis process is triggered, and the root cause analysis is re-performed using the latest data to generate a new decision flow.

[0090] Please see Figure 2 The present invention provides a technical solution: a hydropower station operation data intelligent management system based on lake-warehouse integration. The system is applied to the above-mentioned hydropower station operation data intelligent management method based on lake-warehouse integration. The system includes a hydropower station data processing module, a hydropower station feature analysis module, a resource allocation module, and an operation monitoring and optimization module.

[0091] The hydropower station data processing module is used to collect multi-source operation data of the hydropower station, and perform data cleaning, anomaly detection and normalization preprocessing to generate standardized time series data, which is stored in the lake-warehouse integrated platform.

[0092] The hydropower station feature analysis module is used to analyze the decision value characteristics of each data point based on the preprocessed data, and to analyze value indicators based on the decision value characteristics.

[0093] The resource allocation module is used to dynamically allocate the system's computing and storage resources based on value indicators;

[0094] The operation monitoring and optimization module is used to perform root cause analysis based on high-value data when the operating status is detected to deviate from the preset optimal range. After the decision data is authorized to be executed, the module guides the hydropower station's operating status back to the optimal range based on the deviation reduction rate.

[0095] The hydropower station data processing module includes a hydropower station data acquisition unit and a hydropower station data integration unit;

[0096] The hydropower station data acquisition unit is used to collect the hydropower station's operating data in real time through a sensor network. The hydropower station's operating data includes water level data, flow rate data, equipment power data, equipment temperature data, and equipment vibration data, forming a raw data stream.

[0097] The hydropower station data integration unit is used to clean the raw data stream, identify and remove outliers using a Z-score-based anomaly detection method, normalize various types of data in the data stream after outlier removal to generate standardized data, store the standardized data in the lake-warehouse integrated platform, and output standardized time series data.

[0098] The hydropower station feature analysis module includes a sub-feature analysis unit and a feature fusion unit;

[0099] The sub-feature analysis unit is used to monitor the spatial correlation, time urgency index, and decision contribution of nodes.

[0100] The feature fusion unit combines spatial correlation, time urgency, and decision contribution indicators, and uses a weighted summation method to analyze value indicators.

[0101] The resource allocation module includes a data partitioning unit, an allocation unit, and a threshold correction unit;

[0102] The data partitioning unit is used to set dynamic classification thresholds based on the statistical distribution of historical value indicators. Data with value indicators greater than or equal to the high value threshold are divided into high value groups, data with value indicators less than or equal to the low value threshold are divided into low value groups, and data with value indicators less than the high value threshold but greater than the low value threshold are divided into medium value groups.

[0103] For high-value groups, the allocation unit allocates an in-memory database at the storage level to ensure millisecond-level read and write access to the data; at the computation level, it calls complex algorithms for processing; the complex algorithms include physical mechanism simulation models based on turbine characteristic curves, generator efficiency spectra, and fluid dynamics equations, and unscented Kalman filters; for low-value groups, it uses an arithmetic mean aggregation method to average every certain number of data points and implements downsampling storage; at the computation level, it employs lightweight algorithms; the lightweight algorithms include moving averages and exponential smoothing;

[0104] The threshold correction unit is used to monitor the system resource utilization rate. It corrects the high-value threshold based on the target resource utilization rate and the real-time resource utilization rate, and corrects the low-value threshold based on the corrected high-value threshold. The difference between the corrected low-value threshold and the original low-value threshold is equal to the difference between the corrected high-value threshold and the original high-value threshold.

[0105] The operation monitoring and optimization module includes a status analysis unit and a deviation analysis unit;

[0106] The status analysis unit is used to continuously monitor the operating status indicators. If the deviation of the operating status indicator exceeds the preset tolerance threshold, it is determined that the operating status has deviated from the optimal range and outputs a status deviation signal. It extracts all high-value data corresponding to the timestamp of the detected deviation; it analyzes the mutual information between each potential influencing factor and the deviation signal based on the high-value data; it marks the potential influencing factor with the largest mutual information value as the root cause of the current operating status deviation and generates an early warning signal.

[0107] The deviation analysis unit is used to extract decision data and send it to the administrator to request authorization. After the decision data is authorized and executed, the actual state is compared with the expected state to calculate the deviation reduction rate. If the deviation reduction rate is less than the minimum improvement threshold, the re-analysis process is triggered to re-perform root cause analysis using the latest data and generate a new decision flow.

[0108] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0109] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for intelligent management of hydropower station operation data based on integrated lake and reservoir system, characterized in that: The method includes the following steps: S1: Collect multi-source operation data of hydropower stations, perform data cleaning, anomaly detection and normalization preprocessing, generate standardized time series data, and store them in the integrated lake and warehouse platform; S2: Based on the preprocessed data, analyze the decision value characteristics of each data point, and analyze the value indicators based on the decision value characteristics. S2 includes the following: S201: Obtain the set of spatial neighbor nodes for each monitoring node in the standardized time series data. Denote the set of spatial neighbor nodes for the i-th monitoring node as N(i). Analyze the feature similarity sim(d) between monitoring node i and its spatial neighbor nodes. i d j ); where d i d represents the running data vector of monitoring node i. j sim(d) represents the running data vector of the j-th monitoring node in the set of spatial neighbor nodes. i d j =exp(-||d) i -d j || 2 / (2σ) 2 )); where ||d i -d j || represents the running data vector d i and running data vector d j The Euclidean distance between them; σ represents the preset bandwidth parameter; exp() represents the exponential function with the natural number base; the spatial correlation S of monitoring node i is analyzed based on the feature similarity between monitoring node i and each of its spatial neighbors. i , ； Where N represents the total number of monitoring nodes in the spatial neighbor node set N(i); w （i，j） w represents the spatial association weight between monitoring node i and monitoring node j. （i，j） =1 / (1+d （i，j） ); where d （i，j） This represents the Euclidean distance between monitoring node i and monitoring node j; S202: Analyze data freshness based on the difference between the current time and the data generation time; denot the data freshness of monitoring node i as TN. i TN i =exp(-λ△t i ); where λ represents the preset attenuation factor; Δt i It represents the difference between the current time and the data generation time; the ratio between the change in value and the maximum change in value is recorded as the data change rate; and a weighted sum of the data freshness and data urgency is generated to produce a time urgency index. S203: Based on the similarity analysis of the decision contribution degree of the monitoring point operation data vector and the historical decision corresponding operation data vector, the decision contribution degree of the monitoring node i is recorded as C i , ； Where M represents the number of monitoring nodes corresponding to historical decisions, d m Let U represent the operational data vector of the monitoring node corresponding to the m-th historical decision, f() represent the cosine similarity function, and U m The quantitative value representing the degree of state improvement corresponding to the m-th historical decision is, in this embodiment, the quantitative value of the degree of state improvement is the power deviation reduction rate. S204: Combining spatial relevance, time urgency, and decision contribution, a weighted summation method is used to analyze value indicators; S3: Dynamically allocate the system's computing and storage resources based on value indicators; S3 includes the following: S301: Based on the statistical distribution of historical value indicators, a dynamic classification threshold is set using the percentile method; data with value indicators greater than or equal to the high value threshold are classified as high value group, data with value indicators less than or equal to the low value threshold are classified as low value group, and data with value indicators less than the high value threshold but greater than the low value threshold are classified as medium value group. S302: For high-value groups, allocate a memory database at the storage level to ensure millisecond-level read and write access to the data; at the computation level, call a complex algorithm for processing; the complex algorithm includes a physical mechanism simulation model based on the turbine characteristic curve, generator efficiency spectrum and fluid dynamics equations, and an unscented Kalman filter; For low-value groups, an arithmetic mean aggregation method is used to average every certain number of data points, and downsampling is implemented for storage; at the computational level, a lightweight algorithm is used; the lightweight algorithm includes moving average and exponential smoothing; S303: Monitor system resource utilization rate, adjust high-value thresholds based on target resource utilization rate and real-time resource utilization rate, V high，new =V high，old +k a (U target -U)+k b ∫(U target -U)dt; where V high，new This indicates the adjustment of the high-value threshold, V high，old Indicates the original high-value threshold; k a and k b U represents the control gain coefficient; target denoted as target resource utilization rate, U represents real-time resource utilization rate; the low value threshold is adjusted based on the high value threshold; the difference between the adjusted low value threshold and the original low value threshold is equal to the difference between the adjusted high value threshold and the original high value threshold; S4: When the operating status is found to deviate from the preset optimal range, root cause analysis is performed based on high-value data. After the decision data is authorized for execution, the operating status of the hydropower station is guided back to the optimal range according to the deviation reduction rate. 2.The intelligent management method of the operation data of the hydropower station based on the lake and reservoir integration according to claim 1, characterized in that: In S1, Includes the following: S101: Real-time acquisition of hydropower station operation data through sensor network. The hydropower station operation data includes water level data, flow rate data, equipment power data, equipment temperature data and equipment vibration data, forming a raw data stream, wherein each data point contains timestamp and spatial location information. S102: Perform data cleaning on the raw data stream, and use a Z-score-based anomaly detection method to identify and remove outliers; After removing outliers, the data in the data stream is normalized to generate standardized data. The standardized data is then stored in the lake-warehouse integrated platform, including a data lake for raw storage and a data warehouse for structured queries, outputting standardized time-series data. 3.The intelligent management method of the operation data of the hydropower station based on the lake and the reservoir integration according to claim 2, characterized in that: S4 includes the following: S401: Continuously monitor the operating status indicators. If the deviation of the operating status indicator exceeds the preset tolerance threshold, it is determined that the operating status has deviated from the optimal range, and a status deviation signal is output. The status deviation signal includes a deviation Boolean value, parameter identifier, deviation amount, and timestamp. Extract all high-value data corresponding to the timestamp of the detected deviation. Analyze the mutual information between each potential influencing factor and the deviation signal based on the high-value data. Mark the potential influencing factor with the largest mutual information value as the root cause of the current operating status deviation and generate an early warning signal. S402: Extract decision data and send it to the administrator to request authorization. After the decision data is authorized and executed, compare the actual state with the expected state to calculate the reduction rate of deviation of the operating status index. The reduction rate of deviation of the operating status index is equal to the ratio of the difference between the deviation of the operating status index before and after the adjustment to the deviation of the operating status index before the adjustment. If the deviation reduction rate is less than the minimum improvement threshold, trigger the re-analysis process, use the latest data to re-perform root cause analysis and generate a new decision flow.

4. The intelligent management system for operation data of a hydropower station based on lake and reservoir integration, which is applied to the intelligent management method for operation data of a hydropower station based on lake and reservoir integration according to any one of claims 1-3, characterized in that, The system includes a hydropower station data processing module, a hydropower station characteristic analysis module, a resource allocation module, and an operation monitoring and optimization module; The hydropower station data processing module is used to collect multi-source operation data of the hydropower station, and perform data cleaning, anomaly detection and normalization preprocessing to generate standardized time series data, which is stored in the lake-warehouse integrated platform. The hydropower station feature analysis module is used to analyze the decision value characteristics of each data point based on the preprocessed data, and to analyze value indicators based on the decision value characteristics. The resource allocation module is used to dynamically allocate the system's computing and storage resources based on value indicators; The operation monitoring and optimization module is used to perform root cause analysis based on high-value data when the operation status is detected to deviate from the preset optimal range. After the decision data is authorized to be executed, the module guides the hydropower station's operation status back to the optimal range based on the deviation reduction rate. 5.The intelligent management system for operation data of a hydropower station based on lake and reservoir integration according to claim 4, characterized in that: The hydropower station data processing module includes a hydropower station data acquisition unit and a hydropower station data integration unit; The hydropower station data acquisition unit is used to collect the hydropower station's operating data in real time through a sensor network. The hydropower station's operating data includes water level data, flow rate data, equipment power data, equipment temperature data, and equipment vibration data, forming a raw data stream. The hydropower station data integration unit is used to clean the raw data stream and use a Z-score-based anomaly detection method to identify and remove outliers. After removing outliers, the data in the data stream is normalized to generate standardized data. The standardized data is then stored in the lake warehouse platform, and standardized time series data is output. 6.The intelligent management system for operation data of a hydropower station based on lake and reservoir integration according to claim 4, characterized in that: The hydropower station feature analysis module includes a sub-feature analysis unit and a feature fusion unit; The sub-feature analysis unit is used to monitor the spatial correlation, time urgency index, and decision contribution of nodes. The feature fusion unit combines spatial correlation, time urgency, and decision contribution indicators, and uses a weighted summation method to analyze value indicators. 7.The intelligent management system for operation data of a hydropower station based on lake and reservoir integration according to claim 4, characterized in that: The resource allocation module includes a data partitioning unit, an allocation unit, and a threshold correction unit; The data segmentation unit is used to set dynamic classification thresholds based on the statistical distribution of historical value indicators and using the percentile method; data with value indicators greater than or equal to the high value threshold are classified into the high value group, data with value indicators less than or equal to the low value threshold are classified into the low value group, and data with value indicators less than the high value threshold but greater than the low value threshold are classified into the medium value group. The allocation unit allocates a memory database at the storage level for high-value groups to ensure millisecond-level read and write access to the data; at the computation level, it calls a complex algorithm for processing; the complex algorithm includes a physical mechanism simulation model based on the turbine characteristic curve, generator efficiency spectrum and fluid dynamics equations, and an unscented Kalman filter; for low-value groups, it uses an arithmetic mean aggregation method to take the average of every number of data points and implements downsampling storage. At the computational level, lightweight algorithms are employed; these lightweight algorithms include moving averages and exponential smoothing. The threshold correction unit is used to monitor the system resource utilization rate, correct the high-value threshold based on the target resource utilization rate and the real-time resource utilization rate, and correct the low-value threshold based on the corrected high-value threshold. The difference between the corrected low-value threshold and the original low-value threshold is equal to the difference between the corrected high-value threshold and the original high-value threshold.

8. The intelligent management system for hydropower station operation data based on lake-reservoir integration as described in claim 4, characterized in that: The operation monitoring and optimization module includes a status analysis unit and a deviation analysis unit; The status analysis unit is used to continuously monitor the operating status indicators. If the deviation of the operating status indicator exceeds the preset tolerance threshold, it determines that the operating status has deviated from the optimal range and outputs a status deviation signal. It extracts all high-value data corresponding to the timestamp of the detected deviation. Based on the high-value data, it analyzes the mutual information between each potential influencing factor and the deviation signal. It marks the potential influencing factor with the largest mutual information value as the root cause of the current operating status deviation and generates an early warning signal. The deviation analysis unit is used to extract decision data and send it to the administrator to request authorization. After the decision data is authorized and executed, the actual state is compared with the expected state to calculate the deviation reduction rate. If the deviation reduction rate is less than the minimum improvement threshold; Trigger the reanalysis process, use the latest data to re-perform root cause analysis and generate a new decision flow.

Images