A method and system for online monitoring of the state of a hydroelectric power station unit

Abstract

The application discloses a kind of hydropower station unit state online monitoring method and system, it is related to state monitoring technical field, comprising: obtaining target hydropower station unit whole life cycle operation data, constructs hydropower station unit operation digital twin model;Real-time monitoring multidimensional operation parameter, generates hydropower station unit multidimensional operation visual atlas;Dynamic coupling hydropower station unit operation digital twin model and multidimensional operation visual atlas, quantitatively assesses hydropower station unit operation trajectory health deviation, generates hydropower station unit abnormal area diagnosis report;For hydropower station unit abnormal area diagnosis report, abnormal area real-time monitoring data is input to hydropower station unit operation digital twin model, generates executable predictive maintenance work order, automatically updates hydropower station unit operation digital twin model.The application improves the overall operation reliability and sustainability of hydropower station unit.

Description

Technical Field

[0001] This invention relates to the field of condition monitoring technology, specifically to a method and system for online condition monitoring of hydropower station units. Background Technology

[0002] In terms of online monitoring systems for hydropower station units, traditional sensor data acquisition and static alarm mechanisms are typically relied upon, lacking dynamic real-time monitoring and multi-dimensional comprehensive analysis. These systems often only provide real-time data for single parameters, failing to effectively identify potential unit faults and their development trends, and reacting slowly to complex operating conditions and sudden failures. Existing systems have limited capabilities for fusing and analyzing multi-dimensional operational data, resulting in an inability to accurately predict the unit's health status and remaining lifespan, and a lack of intelligent support for maintenance decisions. These shortcomings make existing technologies inefficient in fault prediction, condition assessment, and maintenance management, failing to fully improve the operational reliability and maintenance accuracy of hydropower station units. Summary of the Invention

[0003] To address the aforementioned technical problems, this paper provides a method and system for online monitoring of the status of hydropower station units. This technical solution solves the problems of low efficiency in fault prediction, status assessment, and maintenance management, which prevents the full improvement of the operational reliability and maintenance accuracy of hydropower station units.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows: Firstly, a method for online monitoring of the status of hydropower station units is provided, including: Acquire the full lifecycle operation data of the target hydropower station units and construct a digital twin model of the hydropower station unit operation; Based on sensors, multi-dimensional operating parameters of hydropower station units are monitored in real time, and a multi-dimensional operation visualization map of hydropower station units is generated. Dynamically couple the digital twin model of hydropower station unit operation with a multi-dimensional operation visualization map to quantitatively assess the health deviation of the hydropower station unit operation trajectory and generate a diagnostic report on abnormal areas of the hydropower station unit. In response to the diagnostic report of abnormal areas of hydropower station units, the enhanced monitoring mode for abnormal areas is activated. The real-time monitoring data of abnormal areas is input into the digital twin model of hydropower station unit operation, generating predictive maintenance work orders that can be executed by hydropower station units, and automatically updating the digital twin model of hydropower station unit operation.

[0005] Preferably, the entire lifecycle operation data of the target hydropower station units is acquired and preprocessed. The full lifecycle operation data includes: Multidimensional operating parameters: temperature, pressure, vibration, voltage, and current; Operating data: Records of start-up, shutdown, power generation, phase adjustment, and load adjustment processes; Equipment files and maintenance records: design parameters, installation records, maintenance reports, component replacement records, and historical ledgers of anomalies and malfunctions; Environmental context data: upstream water level, downstream water level, and water temperature; Based on the known multi-physics coupling mechanism of the turbine, generator and bearing of the target hydropower station, physical models of each subsystem are established. The physical models of each subsystem include: a thermodynamic model, a vibration model, and a hydraulic model.

[0006] Preferably, based on the pre-processed full life cycle operation data of the target hydropower station unit, the data of the absolute health period of the hydropower station unit is screened according to the operation condition data, equipment files and maintenance records; Based on the data from the absolute health period of the hydropower station units, the data was preprocessed and divided into training and validation sets. Based on the training set of data from the absolute health period of hydropower station units, K Gaussian distribution numbers are fitted, and the parameters of each Gaussian distribution are randomly initialized. The Gaussian distribution parameters include: mean, covariance matrix, and weights for each Gaussian distribution. Using Bayes' theorem, calculate the Gaussian distribution probability of data points in the training set of data during the absolute health period of hydropower station units; Based on the Gaussian distribution probability, the Gaussian distribution parameters are updated, and the above process is repeated until the Gaussian distribution parameters converge, thus constructing a data-driven health status assessment model for hydropower station units. For new data points in the full life cycle operation data of the target hydropower station unit, calculate the probability density under the entire Gaussian distribution, set a preset threshold, and determine whether the data point is abnormal. The preset threshold is taken as an example of the 1st percentile or 2.5th percentile of the log-likelihood value of the training set; By integrating the physical models of each subsystem with the data-driven health status assessment model of the hydropower station units, a digital twin model of hydropower station unit operation is constructed.

[0007] The preferred probability density function under a Gaussian distribution is given by the following formula:

[0008] in, The probability density function is a Gaussian distribution. This is a data point for the entire lifecycle operation data center of the target hydropower station units. Let K be the parameter set of the entire Gaussian mixture model, K be the Gaussian distribution number, and k be a constant. For the operating condition vector, Let be the mixture weight function of the k-th Gaussian distribution. Let be the probability density function of the k-th multivariate Gaussian distribution. Let be the mean vector of the k-th Gaussian distribution. Let be the covariance matrix of the k-th Gaussian distribution.

[0009] Preferably, based on sensors, multi-dimensional operating parameters of hydropower station units are monitored in real time, timestamps are aligned, data preprocessing is performed, and time-series data of multi-dimensional operating parameters of hydropower station units are established. Using an LSTM (Long Short-Term Memory) network, the current multi-dimensional operating parameters of the hydropower station unit and the hidden state of the multi-dimensional operating parameters of the hydropower station unit at the previous moment are used as inputs. A sigmoid function is set as the forget gate, and the dynamic deviation between the current multi-dimensional operating parameters of the hydropower station unit and the preset target parameters is used as the input gate for regulation. Combined with the cell state update mechanism, the trend of the multi-dimensional operating parameters of the hydropower station unit over time is captured, and a multi-dimensional operating history trend map of the hydropower station unit is constructed. Based on the time series data of multi-dimensional operating parameters of hydropower station units, the parallel coordinate method is used to map the multi-dimensional operating parameters of hydropower station units onto a two-dimensional plane to generate a parallel coordinate map of the multi-dimensional operating status of hydropower station units. Based on the time series data of multi-dimensional operating parameters of hydropower station units, the sampling frequency of multi-dimensional operating parameters of hydropower station units is calculated according to the Nyquist theorem, and the time-domain vibration signal of multi-dimensional operating parameters of hydropower station units is obtained. Based on the time-domain vibration signal of multi-dimensional operating parameters of hydropower station units, signal preprocessing is performed; The FFT (Fast Fourier Transform) was used to convert the time-domain vibration signal of the multi-dimensional operating parameters of the hydropower station unit into the frequency-domain representation of the multi-dimensional operating parameters of the hydropower station unit. Calculate the frequency domain amplitude spectrum of multi-dimensional operating parameters of hydropower station units, extract the frequency of maximum amplitude and frequency centroid, construct the frequency domain feature matrix of multi-dimensional operating parameters of hydropower station units, extract the frequency domain features of multi-dimensional operating parameters of hydropower station units, and construct the multi-dimensional operating spectrum of hydropower station units. Integrate the multi-dimensional operation history trend chart, parallel coordinate chart and spectrum chart of hydropower station units to generate a multi-dimensional operation visualization map of hydropower station units.

[0010] Preferably, a health benchmark map library for hydropower station units is constructed based on the absolute health period data of the hydropower station units and combined with the multi-dimensional operation visualization map of the hydropower station units. Based on real-time monitoring of multi-dimensional operating parameters of hydropower station units, input into the digital twin model of hydropower station unit operation, and generate the expected health benchmark map of hydropower station units; For the multi-dimensional operation visualization map and the expected health benchmark map of hydropower station units, the structural similarity index algorithm is used to compare the differences in brightness, contrast and structural features to quantitatively assess the health deviation of the hydropower station unit's operation trajectory.

[0011] Preferably, the model uses the health maps and labeled abnormal maps in the health benchmark map library of hydropower station units as input, and cross-entropy as the loss function to train the fault classification model of hydropower station units, and sets the discrimination threshold to realize the automatic identification of abnormal maps of hydropower station units. The discrimination threshold is set differently for different operating conditions; The identified hydropower station unit anomaly map is input into the hydropower station unit fault classification model. The forward propagation method is used to obtain the fault classification prediction probability of the hydropower station unit. Global average pooling is performed to obtain the distribution weight of the hydropower station unit anomaly area and generate a low-resolution heat map of the hydropower station unit anomaly area. Using the upsampling method, the low-resolution heat map of the abnormal area of ​​the hydropower station unit is reconstructed into a high-resolution heat map, and a diagnostic report of the abnormal area of ​​the hydropower station unit is generated.

[0012] Preferably, for the diagnostic report of abnormal areas of hydropower station units, abnormal areas of hydropower station units are identified, and the sampling frequency of abnormal area monitoring is adaptively increased; Based on real-time monitoring data of abnormal areas, wavelet transform is used to extract the time-frequency characteristics of real-time monitoring signals of abnormal areas and obtain the instantaneous frequency evolution of the state of abnormal areas. Based on real-time monitoring data of abnormal areas and changes in the failure modes of hydropower station units, the monitoring points in abnormal areas are dynamically adjusted. Real-time monitoring data of abnormal areas are input into the digital twin model of hydropower station unit operation. Particle filtering is used to dynamically update the prediction of hydropower station unit operation status. The model hyperparameters are optimized through cross-validation to realize the prediction of the remaining useful life of hydropower station units. Based on the prediction of the remaining useful life of hydropower station units, the system automatically generates predictive maintenance work orders for hydropower station units, which clearly define maintenance tasks, time nodes and required resources, providing clear action guidelines for operation and maintenance personnel. During the execution of maintenance work orders, the system collects actual maintenance data and feeds it back to the digital twin model of hydropower station unit operation in a closed loop, so as to realize automatic calibration and continuous updating of model parameters.

[0013] Secondly, an online monitoring system for the status of hydropower station units is also provided, which implements a method for online monitoring of the status of hydropower station units, including: The system includes a data acquisition module, a visualization map module, an anomaly area diagnosis report module, and a data feedback module. The data acquisition module is used to acquire the full life cycle operation data of the target hydropower station unit and construct a digital twin model of the hydropower station unit operation. The visualization map is used to monitor multi-dimensional operating parameters of hydropower station units in real time based on sensors, and generate a multi-dimensional operating visualization map of hydropower station units. The abnormal area diagnosis report module is electrically connected to the visualization map module and the data acquisition module. It is used to dynamically couple the digital twin model of hydropower station unit operation with the multi-dimensional operation visualization map, quantitatively evaluate the health deviation of the hydropower station unit operation trajectory, and generate an abnormal area diagnosis report of the hydropower station unit. The data feedback module is electrically connected to the abnormal area diagnosis report module. It is used to activate the abnormal area enhanced monitoring mode for the abnormal area diagnosis report of the hydropower station unit, input the real-time monitoring data of the abnormal area into the digital twin model of the hydropower station unit operation, generate the predictive maintenance work order that can be executed by the hydropower station unit, and automatically update the digital twin model of the hydropower station unit operation.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention proposes an online monitoring method and system for the status of hydropower station units. It introduces digital twin technology and dynamic coupling with a multi-dimensional operational visualization map, significantly improving the real-time performance and comprehensive analysis capabilities of the monitoring system. By quantitatively assessing the health deviation of the operating trajectory and generating accurate diagnostic reports for abnormal areas, it can capture potential faults in real time, improving the accuracy and response speed of fault diagnosis. The solution automatically generates predictive maintenance work orders and dynamically updates the digital twin model, optimizing operation and maintenance decisions, achieving accurate prediction of remaining life, and intelligent maintenance. This reduces equipment downtime and maintenance costs, and improves the overall operational reliability and sustainability of the hydropower station units. Attached Figure Description

[0015] Figure 1 A flowchart of an online monitoring method for the status of hydropower station units; Figure 2 This is a framework diagram of an online monitoring system for the status of hydropower station units. Detailed Implementation

[0016] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0017] Reference Figure 1 As shown, an embodiment of the present invention provides a method for online monitoring of the status of hydropower station units, including: S1. Obtain the full life cycle operation data of the target hydropower station units and construct a digital twin model of the hydropower station unit operation; Step S1 includes the following: Acquire the full lifecycle operation data of the target hydropower station units and perform data preprocessing; The full lifecycle operation data includes: Multidimensional operating parameters: temperature, pressure, vibration, voltage, and current; Operating data: Records of start-up, shutdown, power generation, phase adjustment, and load adjustment processes; Equipment files and maintenance records: design parameters, installation records, maintenance reports, component replacement records, and historical ledgers of anomalies and malfunctions; Environmental context data: upstream water level, downstream water level, and water temperature; Based on the known multi-physics coupling mechanism of the turbine, generator and bearing of the target hydropower station, physical models of each subsystem are established. The physical models of each subsystem include: a thermodynamic model, a vibration model, and a hydraulic model.

[0018] Step S1 also includes the following: Based on the pre-processed full life cycle operation data of the target hydropower station unit, and according to the operating condition data, equipment files and maintenance records, the data of the absolute health period of the hydropower station unit is screened. Based on the data from the absolute health period of the hydropower station units, the data was preprocessed and divided into training and validation sets. Based on the training set of data from the absolute health period of hydropower station units, K Gaussian distribution numbers are fitted, and the parameters of each Gaussian distribution are randomly initialized. The Gaussian distribution parameters include: mean, covariance matrix, and weights for each Gaussian distribution. Using Bayes' theorem, calculate the Gaussian distribution probability of data points in the training set of data during the absolute health period of hydropower station units; Based on the Gaussian distribution probability, the Gaussian distribution parameters are updated, and the above process is repeated until the Gaussian distribution parameters converge, thus constructing a data-driven health status assessment model for hydropower station units. For new data points in the full life cycle operation data of the target hydropower station unit, calculate the probability density under the entire Gaussian distribution, set a preset threshold, and determine whether the data point is abnormal. The probability density function under the Gaussian distribution is given by the following formula:

[0019] in, The probability density function is a Gaussian distribution. This is a data point for the entire lifecycle operation data center of the target hydropower station units. Let K be the parameter set of the entire Gaussian mixture model, K be the Gaussian distribution number, and k be a constant. For the operating condition vector, Let be the mixture weight function of the k-th Gaussian distribution. Let be the probability density function of the k-th multivariate Gaussian distribution. Let be the mean vector of the k-th Gaussian distribution. Let be the covariance matrix of the k-th Gaussian distribution; The preset threshold is taken as an example of the 1st percentile or 2.5th percentile of the log-likelihood value of the training set; By integrating the physical models of each subsystem with the data-driven health status assessment model of the hydropower station units, a digital twin model of hydropower station unit operation is constructed.

[0020] When using it, refer to the content of step S1 above: Existing domestic and international technologies for health assessment of hydropower station units typically rely on static fault diagnosis models or single data sources, lacking a comprehensive approach that considers operational data throughout the entire equipment lifecycle. Current technologies often use only single-dimensional parameters, such as temperature and pressure, and generally neglect the deep integration and dynamic feedback of multi-dimensional data, including operating conditions, equipment records, and maintenance logs. The accuracy of traditional models is limited by manually set thresholds and rules, making them unable to adapt to real-time changes in equipment status and struggling to handle the complexity of multi-physics coupling. Existing technologies have significant limitations in the accuracy and real-time performance of fault prediction and health assessment. This step... By integrating the entire lifecycle operation data of hydropower station units and combining it with multi-dimensional operating parameters, environmental data, equipment records, and other information, a digital twin model is constructed, enabling a more comprehensive and accurate health status assessment. Through probability calculation based on Gaussian distribution and Bayesian methods, the equipment health model is dynamically fitted, and the fusion of physical models and data-driven models can more accurately determine the equipment status and detect anomalies in a timely manner, thereby improving the accuracy and response speed of fault prediction. The construction of the digital twin model also provides strong technical support for remote monitoring, intelligent operation and maintenance, and adaptive optimization of hydropower station units, significantly improving the operating efficiency and safety of hydropower stations.

[0021] S2. Based on sensors, monitor multi-dimensional operating parameters of hydropower station units in real time and generate a multi-dimensional operation visualization map of hydropower station units; the multi-dimensional operating parameters include: temperature, pressure, vibration, voltage and current; Step S2 includes the following: Based on sensors, multi-dimensional operating parameters of hydropower station units are monitored in real time, timestamps are aligned, data preprocessing is performed, and time-series data of multi-dimensional operating parameters of hydropower station units are established. Using an LSTM (Long Short-Term Memory) network, the current multi-dimensional operating parameters of the hydropower station unit and the hidden state of the multi-dimensional operating parameters of the hydropower station unit at the previous moment are used as inputs. A sigmoid function is set as the forget gate, and the dynamic deviation between the current multi-dimensional operating parameters of the hydropower station unit and the preset target parameters is used as the input gate for regulation. Combined with the cell state update mechanism, the trend of the multi-dimensional operating parameters of the hydropower station unit over time is captured, and a multi-dimensional operating history trend map of the hydropower station unit is constructed. Based on the time series data of multi-dimensional operating parameters of hydropower station units, the parallel coordinate method is used to map the multi-dimensional operating parameters of hydropower station units onto a two-dimensional plane to generate a parallel coordinate map of the multi-dimensional operating status of hydropower station units. Based on the time series data of multi-dimensional operating parameters of hydropower station units, the sampling frequency of multi-dimensional operating parameters of hydropower station units is calculated according to the Nyquist theorem, and the time-domain vibration signal of multi-dimensional operating parameters of hydropower station units is obtained. Based on the time-domain vibration signal of multi-dimensional operating parameters of hydropower station units, signal preprocessing is performed; The FFT (Fast Fourier Transform) was used to convert the time-domain vibration signal of the multi-dimensional operating parameters of the hydropower station unit into the frequency-domain representation of the multi-dimensional operating parameters of the hydropower station unit. Calculate the frequency domain amplitude spectrum of multi-dimensional operating parameters of hydropower station units, extract the frequency of maximum amplitude and frequency centroid, construct the frequency domain feature matrix of multi-dimensional operating parameters of hydropower station units, extract the frequency domain features of multi-dimensional operating parameters of hydropower station units, and construct the multi-dimensional operating spectrum of hydropower station units. Integrate the multi-dimensional operation history trend chart, parallel coordinate chart and spectrum chart of hydropower station units to generate a multi-dimensional operation visualization map of hydropower station units.

[0022] When using it, refer to the content of step S2 above: Existing domestic and international technologies for monitoring and visualizing multi-dimensional operating parameters of hydropower station units typically rely on traditional sensor data acquisition and static chart displays, failing to fully utilize advanced data analysis and deep learning technologies. Traditional methods often only capture real-time data for single parameters and lack in-depth time-series analysis of dynamic trends, making it difficult to capture long-term trends or sudden changes in equipment status. This paper, by combining advanced technologies such as LSTM (Long Short-Term Memory) networks, parallel coordinate graphs, and FFT (Fast Fourier Transform), can more comprehensively and accurately capture the dynamic trends of multi-dimensional operating parameters of hydropower station units. By generating multi-dimensional operational visualization maps, it provides a clearer and more intuitive display of equipment status. Through the integration and analysis of historical trend charts and spectrum charts, it not only improves the ability to identify potential equipment faults but also enables real-time monitoring and prediction of operating status, providing strong data support for intelligent operation and maintenance of hydropower stations and effectively improving fault prediction accuracy and response speed.

[0023] S3. Dynamically couple the digital twin model of hydropower station unit operation with a multi-dimensional operation visualization map to quantitatively assess the health deviation of the hydropower station unit operation trajectory and generate a diagnostic report on abnormal areas of the hydropower station unit. Step S3 includes the following: Based on the data of the absolute health period of hydropower station units, and combined with the multi-dimensional operation visualization map of hydropower station units, a health benchmark map library of hydropower station units is constructed. Based on real-time monitoring of multi-dimensional operating parameters of hydropower station units, input into the digital twin model of hydropower station unit operation, and generate the expected health benchmark map of hydropower station units; For the multi-dimensional operation visualization map and the expected health benchmark map of hydropower station units, the structural similarity index algorithm is used to compare the differences in brightness, contrast and structural features to quantitatively assess the health deviation of the hydropower station unit's operation trajectory.

[0024] Step S3 also includes the following: Using the health maps and labeled abnormal maps in the health benchmark map library of hydropower station units as input, and cross-entropy as the loss function, a fault classification model for hydropower station units is trained, and a discrimination threshold is set to realize the automatic identification of abnormal maps of hydropower station units. The discrimination threshold is set differently for different operating conditions; The identified hydropower station unit anomaly map is input into the hydropower station unit fault classification model. The forward propagation method is used to obtain the fault classification prediction probability of the hydropower station unit. Global average pooling is performed to obtain the distribution weight of the hydropower station unit anomaly area and generate a low-resolution heat map of the hydropower station unit anomaly area. Using the upsampling method, the low-resolution heat map of the abnormal area of ​​the hydropower station unit is reconstructed into a high-resolution heat map, and a diagnostic report of the abnormal area of ​​the hydropower station unit is generated.

[0025] When using it, refer to the content of step S3 above: Existing domestic and international technologies for health assessment and fault diagnosis of hydropower station units typically rely on traditional static models and single data sources, lacking dynamic and multi-dimensional comprehensive analysis. They often base comparisons on fixed health benchmarks, failing to consider the real-time operating conditions and variations under different working environments of the hydropower station units, resulting in limited quantitative assessments of equipment health deviations. Current technologies are weak in the fusion and dynamic monitoring of multi-dimensional operating parameters, making it difficult to quickly identify and accurately locate potential fault areas under real-time data changes. This step achieves a more accurate and real-time quantitative assessment of health deviations by dynamically coupling a digital twin model of the hydropower station unit with a multi-dimensional operational visualization map, combined with a structural similarity index algorithm and a fault classification model. By generating high-resolution heat maps of abnormal areas, the accuracy of fault diagnosis is improved, enabling real-time identification of equipment anomalies and providing precise diagnostic reports. This enhances the intelligent operation and maintenance capabilities of the hydropower station, improves the accuracy and response speed of equipment health management, and more effectively supports fault prediction and maintenance decisions.

[0026] S4. For the abnormal area diagnosis report of the hydropower station unit, start the abnormal area enhanced monitoring mode, input the real-time monitoring data of the abnormal area into the hydropower station unit operation digital twin model, generate the hydropower station unit executable predictive maintenance work order, and automatically update the hydropower station unit operation digital twin model. Step S4 includes the following: For the diagnostic report of abnormal areas of hydropower station units, identify abnormal areas of hydropower station units and adaptively increase the monitoring sampling frequency of abnormal areas; Based on real-time monitoring data of abnormal areas, wavelet transform is used to extract the time-frequency characteristics of real-time monitoring signals of abnormal areas and obtain the instantaneous frequency evolution of the state of abnormal areas. Based on real-time monitoring data of abnormal areas and changes in the failure modes of hydropower station units, the monitoring points in abnormal areas are dynamically adjusted. Real-time monitoring data of abnormal areas are input into the digital twin model of hydropower station unit operation. Particle filtering is used to dynamically update the prediction of hydropower station unit operation status. The model hyperparameters are optimized through cross-validation to realize the prediction of the remaining useful life of hydropower station units. Based on the prediction of the remaining useful life of hydropower station units, the system automatically generates predictive maintenance work orders for hydropower station units, which clearly define maintenance tasks, time nodes and required resources, providing clear action guidelines for operation and maintenance personnel. During the execution of maintenance work orders, the system collects actual maintenance data and feeds it back to the digital twin model of hydropower station unit operation in a closed loop, so as to realize automatic calibration and continuous updating of model parameters.

[0027] When using it, refer to the content of step S4 above: Existing domestic and international technologies for monitoring and maintaining abnormal areas of hydropower station units largely rely on traditional fault diagnosis methods and periodic inspection models. These methods typically struggle to achieve real-time monitoring and dynamic adjustment of abnormal areas, and lack refined and intelligent support for fault early warning and maintenance decision-making. They also fail to flexibly respond to sudden failures or changing working environments, and lack accurate prediction of equipment remaining life and automated maintenance scheduling, resulting in low equipment operating efficiency and high maintenance costs. This approach improves the accuracy of fault diagnosis and prediction for hydropower station units by dynamically adjusting the monitoring frequency and sampling points of abnormal areas, combined with wavelet transform and particle filtering methods, to extract the time-frequency characteristics of abnormal areas in real time and update the operational digital twin model. By automatically generating predictive maintenance work orders and providing clear action guidelines, it achieves accurate remaining life prediction and intelligent maintenance planning, optimizes the allocation of operation and maintenance resources, and reduces equipment downtime and maintenance costs. This not only improves maintenance efficiency but also enhances the reliability and sustainability of hydropower station unit operation, promoting the development of intelligent operation and maintenance towards greater efficiency and refinement.

[0028] Reference Figure 2As shown, one embodiment of the present invention also provides an online monitoring system for the status of hydropower station units, comprising: The system includes a data acquisition module, a visualization map module, an anomaly area diagnosis report module, and a data feedback module. The data acquisition module is used to acquire the full life cycle operation data of the target hydropower station unit and construct a digital twin model of the hydropower station unit operation. The visualization map is used to monitor multi-dimensional operating parameters of hydropower station units in real time based on sensors, and generate a multi-dimensional operating visualization map of hydropower station units. The abnormal area diagnosis report module is electrically connected to the visualization map module and the data acquisition module. It is used to dynamically couple the digital twin model of hydropower station unit operation with the multi-dimensional operation visualization map, quantitatively evaluate the health deviation of the hydropower station unit operation trajectory, and generate an abnormal area diagnosis report of the hydropower station unit. The data feedback module is electrically connected to the abnormal area diagnosis report module. It is used to activate the abnormal area enhanced monitoring mode for the abnormal area diagnosis report of the hydropower station unit, input the real-time monitoring data of the abnormal area into the digital twin model of the hydropower station unit operation, generate the predictive maintenance work order that can be executed by the hydropower station unit, and automatically update the digital twin model of the hydropower station unit operation.

[0029] The foregoing has shown and described 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. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A method for online monitoring of the status of hydropower station units, characterized in that, include: S1. Obtain the full life cycle operation data of the target hydropower station units and construct a digital twin model of the hydropower station unit operation; S2. Based on sensors, monitor multi-dimensional operating parameters of hydropower station units in real time and generate a multi-dimensional operation visualization map of hydropower station units. S3. Dynamically couple the digital twin model of hydropower station unit operation with a multi-dimensional operation visualization map to quantitatively assess the health deviation of the hydropower station unit operation trajectory and generate a diagnostic report on abnormal areas of the hydropower station unit. S4. For the abnormal area diagnosis report of the hydropower station unit, activate the abnormal area enhanced monitoring mode, input the real-time monitoring data of the abnormal area into the hydropower station unit operation digital twin model, generate the hydropower station unit executable predictive maintenance work order, and automatically update the hydropower station unit operation digital twin model.

2. The method for online monitoring of the status of hydropower station units according to claim 1, characterized in that, S1 includes: Acquire the full lifecycle operation data of the target hydropower station units and perform data preprocessing; The full lifecycle operation data includes: Multidimensional operating parameters: temperature, pressure, vibration, voltage, and current; Operating data: Records of start-up, shutdown, power generation, phase adjustment, and load adjustment processes; Equipment files and maintenance records: design parameters, installation records, maintenance reports, component replacement records, and historical ledgers of anomalies and malfunctions; Environmental context data: upstream water level, downstream water level, and water temperature; Based on the known multi-physics coupling mechanism of the turbine, generator and bearing of the target hydropower station, physical models of each subsystem are established. The physical models of each subsystem include: a thermodynamic model, a vibration model, and a hydraulic model.

3. The method for online monitoring of the status of hydropower station units according to claim 2, characterized in that, S1 further includes: Based on the pre-processed full life cycle operation data of the target hydropower station unit, and according to the operating condition data, equipment files and maintenance records, the data of the absolute health period of the hydropower station unit is screened. Based on the data from the absolute health period of the hydropower station units, the data was preprocessed and divided into training and validation sets. Based on the training set of data from the absolute health period of hydropower station units, K Gaussian distribution numbers are fitted, and the parameters of each Gaussian distribution are randomly initialized. The Gaussian distribution parameters include: mean, covariance matrix, and weights for each Gaussian distribution. Using Bayes' theorem, calculate the Gaussian distribution probability of data points in the training set of data during the absolute health period of hydropower station units; Based on the Gaussian distribution probability, the Gaussian distribution parameters are updated, and the above process is repeated until the Gaussian distribution parameters converge, thus constructing a data-driven health status assessment model for hydropower station units. For new data points in the full life cycle operation data of the target hydropower station unit, calculate the probability density under the entire Gaussian distribution, set a preset threshold, and determine whether the data point is abnormal. The preset threshold is taken as an example of the 1st percentile or 2.5th percentile of the log-likelihood value of the training set; By integrating the physical models of each subsystem with the data-driven health status assessment model of the hydropower station units, a digital twin model of hydropower station unit operation is constructed.

4. The method for online monitoring of the status of hydropower station units according to claim 3, characterized in that, The probability density function under the Gaussian distribution is given by the following formula: in, The probability density function is a Gaussian distribution. This is a data point for the entire lifecycle operation data center of the target hydropower station units. Let K be the parameter set of the entire Gaussian mixture model, K be the Gaussian distribution number, and k be a constant. For the operating condition vector, Let be the mixture weight function of the k-th Gaussian distribution. Let be the probability density function of the k-th multivariate Gaussian distribution. Let be the mean vector of the k-th Gaussian distribution. Let be the covariance matrix of the k-th Gaussian distribution.

5. The method for online monitoring of the status of hydropower station units according to claim 1, characterized in that, S2 includes: Based on sensors, multi-dimensional operating parameters of hydropower station units are monitored in real time, timestamps are aligned, data preprocessing is performed, and time-series data of multi-dimensional operating parameters of hydropower station units are established. Using an LSTM (Long Short-Term Memory) network, the current multi-dimensional operating parameters of the hydropower station units and the hidden state of the multi-dimensional operating parameters of the hydropower station units at the previous moment are used as inputs. A sigmoid function is set as the forget gate, and the dynamic deviation between the current multi-dimensional operating parameters of the hydropower station units and the preset target parameters is used as the input gate for regulation. Combined with the cell state update mechanism, the trend of the multi-dimensional operating parameters of the hydropower station units over time is captured, and a multi-dimensional operating history trend map of the hydropower station units is constructed. Based on the time series data of multi-dimensional operating parameters of hydropower station units, the parallel coordinate method is used to map the multi-dimensional operating parameters of hydropower station units onto a two-dimensional plane to generate a parallel coordinate map of the multi-dimensional operating status of hydropower station units. Based on the time series data of multi-dimensional operating parameters of hydropower station units, the sampling frequency of multi-dimensional operating parameters of hydropower station units is calculated according to the Nyquist theorem, and the time-domain vibration signal of multi-dimensional operating parameters of hydropower station units is obtained. Based on the time-domain vibration signal of multi-dimensional operating parameters of hydropower station units, signal preprocessing is performed; The FFT (Fast Fourier Transform) was used to convert the time-domain vibration signal of the multi-dimensional operating parameters of the hydropower station unit into the frequency-domain representation of the multi-dimensional operating parameters of the hydropower station unit. Calculate the frequency domain amplitude spectrum of multi-dimensional operating parameters of hydropower station units, extract the frequency of maximum amplitude and frequency centroid, construct the frequency domain feature matrix of multi-dimensional operating parameters of hydropower station units, extract the frequency domain features of multi-dimensional operating parameters of hydropower station units, and construct the multi-dimensional operating spectrum of hydropower station units. The system integrates multi-dimensional historical trend charts, parallel coordinate charts, and spectrum charts of hydropower station units to generate a multi-dimensional operational visualization map of hydropower station units.

6. The method for online monitoring of the status of hydropower station units according to claim 5, characterized in that, S3 includes: Based on the data of the absolute health period of hydropower station units, and combined with the multi-dimensional operation visualization map of hydropower station units, a health benchmark map library of hydropower station units is constructed. Based on real-time monitoring of multi-dimensional operating parameters of hydropower station units, input into the digital twin model of hydropower station unit operation, and generate the expected health benchmark map of hydropower station units; For the multi-dimensional operation visualization map and the expected health benchmark map of hydropower station units, the structural similarity index algorithm is used to compare the differences in brightness, contrast and structural features to quantitatively assess the health deviation of the hydropower station unit's operation trajectory.

7. The method for online monitoring of the status of hydropower station units according to claim 6, characterized in that, S3 further includes: Using the health maps and labeled abnormal maps in the health benchmark map library of hydropower station units as input, and cross-entropy as the loss function, a fault classification model for hydropower station units is trained, and a discrimination threshold is set to realize the automatic identification of abnormal maps of hydropower station units. The discrimination threshold is set differently for different operating conditions; The identified hydropower station unit anomaly map is input into the hydropower station unit fault classification model. The forward propagation method is used to obtain the fault classification prediction probability of the hydropower station unit. Global average pooling is performed to obtain the distribution weight of the hydropower station unit anomaly area and generate a low-resolution heat map of the hydropower station unit anomaly area. Using the upsampling method, the low-resolution heat map of the abnormal area of ​​the hydropower station unit is reconstructed into a high-resolution heat map, and a diagnostic report of the abnormal area of ​​the hydropower station unit is generated.

8. The method for online monitoring of the status of hydropower station units according to claim 7, characterized in that, S4 includes: For the diagnostic report of abnormal areas of hydropower station units, identify abnormal areas of hydropower station units and adaptively increase the monitoring sampling frequency of abnormal areas; Based on real-time monitoring data of abnormal areas, wavelet transform is used to extract the time-frequency characteristics of real-time monitoring signals of abnormal areas and obtain the instantaneous frequency evolution of the state of abnormal areas. Based on real-time monitoring data of abnormal areas and changes in the failure modes of hydropower station units, the monitoring points in abnormal areas are dynamically adjusted. Real-time monitoring data of abnormal areas are input into the digital twin model of hydropower station unit operation. Particle filtering is used to dynamically update the prediction of hydropower station unit operation status. The model hyperparameters are optimized through cross-validation to realize the prediction of the remaining useful life of hydropower station units. Based on the prediction of the remaining useful life of hydropower station units, the system automatically generates predictive maintenance work orders for hydropower station units, which clearly define maintenance tasks, time nodes and required resources, providing clear action guidelines for operation and maintenance personnel. During the execution of maintenance work orders, the system collects actual maintenance data and feeds it back to the digital twin model of the hydropower station unit operation in a closed loop, so as to realize automatic calibration and continuous updating of model parameters.

9. An online monitoring system for the status of hydropower station units, characterized in that, The method for online monitoring of the status of hydropower station units according to any one of claims 1-8, wherein the online monitoring system for the status of hydropower station units includes a data acquisition module, a visualization map module, an abnormal area diagnosis report module, and a data feedback module; The data acquisition module is used to acquire the full life cycle operation data of the target hydropower station unit and construct a digital twin model of the hydropower station unit operation. The visualization map module is used to monitor the multi-dimensional operating parameters of hydropower station units in real time based on sensors and generate a multi-dimensional operation visualization map of hydropower station units. The abnormal area diagnosis report module is electrically connected to the visualization map module and the data acquisition module. It is used to dynamically couple the digital twin model of hydropower station unit operation with the multi-dimensional operation visualization map, quantitatively evaluate the health deviation of the hydropower station unit operation trajectory, and generate an abnormal area diagnosis report of the hydropower station unit. The data feedback module is electrically connected to the abnormal area diagnosis report module. It is used to activate the abnormal area enhanced monitoring mode for the abnormal area diagnosis report of the hydropower station unit, input the real-time monitoring data of the abnormal area into the digital twin model of the hydropower station unit operation, generate the predictive maintenance work order that can be executed by the hydropower station unit, and automatically update the digital twin model of the hydropower station unit operation.

Images