A system and method for abnormal monitoring of a hydro-generating unit

Abstract

The present invention relates to the technical field of abnormal monitoring of hydro-generating units, and in particular to a system and method for abnormal monitoring of hydro-generating units, including a sensor module, an edge computing node, a cloud analysis platform, and a dispatching decision-making center. The sensor module comprehensively collects the operation and hydrological data of the unit, the edge computing node performs data preprocessing, the cloud analysis platform uses technologies such as multi-source data fusion, dynamic early warning model library, and adaptive threshold adjustment to achieve accurate early warning, the dispatching decision-making center optimizes the dispatching decision through a resource matching engine and an emergency plan generation module, and the industrial communication network ensures reliable data transmission. In addition, a visual interaction interface and a status evaluation report generation module are also provided. The present invention realizes comprehensive, accurate monitoring and intelligent early warning of hydro-generating units, effectively improves the efficiency of fault handling, reduces the probability of false alarms and missed alarms, optimizes resource allocation, enhances operation safety, and provides a strong guarantee for the stable operation of hydro-generating units.

Description

Technical Field

[0001] The present invention relates to the technical field of abnormal monitoring of hydro-generator units, and specifically provides a system and method for abnormal monitoring of hydro-generator units. Background Art

[0002] Hydro-generator units play a core role in the field of hydropower generation and are crucial for ensuring stable power supply. With the increasing global demand for clean energy, hydropower generation has become increasingly important in the energy industry due to its clean and renewable characteristics. However, the operating environment of hydro-generator units is extremely complex.

[0003] From the perspective of the unit itself, its key components are under high load operation for a long time. Components such as the main shaft and bearings are constantly subjected to mechanical stress, which makes it easy for key parameters such as vibration, temperature, and pressure to show abnormalities. In addition, the hydrological conditions around the hydraulic structure are unpredictable. Large fluctuations in water level and unstable water flow velocity will indirectly affect the stable operation of the unit. Moreover, the grid load is in dynamic change, and the unit needs to frequently adjust the output power, which undoubtedly increases the complexity of operation.

[0004] Traditional monitoring methods for hydro-generator units have obvious shortcomings. Early monitoring systems only rely on basic sensors to monitor single parameters and cannot comprehensively grasp the operating state of the unit. In the data processing link, due to the lack of powerful edge computing and data analysis capabilities, it is difficult to process a large amount of sensor data in real time and accurately when facing massive sensor data, resulting in potential fault hazards being difficult to detect in a timely manner. In terms of fault warning, traditional systems mostly adopt a fixed threshold alarm mode and cannot flexibly adjust according to the equipment operation duration, maintenance records, and real-time working conditions, resulting in frequent false alarms and missed alarms. Once a fault occurs, due to the lack of an efficient resource matching and emergency plan generation mechanism, the fault handling efficiency is greatly reduced, and the maintenance cost also soars. Therefore, a system and method for abnormal monitoring of hydro-generator units are proposed to address the above problems. Summary of the Invention

[0005] The purpose of the present invention is to provide a system and method for abnormal monitoring of hydro-generator units to solve the problems raised in the above background art.

[0006] To achieve the above purpose, the present invention provides the following technical solutions:

[0007] A system for abnormal monitoring of hydro-generator units, comprising:

[0008] A sensor module: including a vibration sensor array, a temperature sensor group, a pressure sensor, and a flow sensor arranged at key parts of the hydro-generator unit body, as well as a hydrological monitoring unit provided in the hydraulic structure;

[0009] Edge computing node: Configured to receive and preprocess sensor data, including a data verification module, a feature extraction module, and a local storage unit;

[0010] Cloud analysis platform: Provided with:

[0011] Multi-source data fusion module: Used to integrate unit operation data, hydrological data, and power grid load data;

[0012] Dynamic early warning model library: Includes a vibration trend prediction sub-model based on improved LSTM, a temperature anomaly classification sub-model driven by a weighted support vector machine, and a fault mode recognition sub-model based on transfer learning;

[0013] Adaptive threshold adjustment module: Dynamically corrects the alarm threshold according to the equipment operation duration and maintenance records;

[0014] Dispatch decision-making center: Includes:

[0015] Resource matching engine: Connects the spare parts inventory database and the personnel qualification database;

[0016] Emergency plan generation module: Configured to generate an emergency plan including equipment shutdown suggestions, spare parts allocation plans, and emergency repair personnel dispatch instructions according to the early warning level;

[0017] Visualization interaction interface: Provides three-dimensional equipment status display and operation confirmation functions;

[0018] Industrial communication network: Adopts the OPCUA protocol to connect each module and is provided with a time-sensitive network channel to ensure the transmission of key data.

[0019] As a preferred solution, the vibration sensor array adopts a spatially distributed layout scheme, including:

[0020] Axially arranged acceleration sensor group: Installed at equal intervals along the main shaft axis, and the distance d between adjacent acceleration sensors is less than one-fourth of the wavelength corresponding to the lowest characteristic frequency of the hydro-generator set;

[0021] Radially arranged composite sensor: Integrates vibration and displacement detection functions and is provided with a temperature compensation unit;

[0022] Phase synchronization acquisition module: Configured to align the clock of the vibration signals at each measurement point, and the adaptive adjustment range of the sampling frequency is 1 kHz - 20 kHz.

[0023] As a preferred solution, the working method of the dynamic early warning model library includes:

[0024] Establishing the equipment health baseline: The improved hybrid wavelet packet decomposition algorithm is used to extract features. The wavelet coefficient of the k-th node in the j-th layer is calculated by multiplying the input discrete-time series signal by the complex conjugate of the modified Morlet wavelet basis function and then multiplying by an exponential decay term. The decay factor is used to adjust the decay speed, and the central position of the fault-sensitive area corresponds to the discrete-time serial number.

[0025] Real-time anomaly detection: The weighted Mahalanobis distance is used to measure the deviation of the real-time feature vector from the benchmark. The weighted Mahalanobis distance is equal to the product of the transposed vector of the real-time feature vector minus the benchmark mean vector and the inverse matrix of the covariance matrix, and then multiplied by a time-varying weight coefficient.

[0026] Multi-model collaborative decision-making: When the alarm confidence level of the sub-model exceeds the threshold, the final alarm probability is calculated by mapping the output probability of the vibration trend prediction sub-model based on the improved LSTM and the probability after combining the other two sub-models through a dynamic weight coefficient and an activation function.

[0027] As a preferred solution, the adaptive threshold adjustment module includes:

[0028] The working condition adaptation unit, where the vibration threshold is calculated by multiplying the benchmark vibration threshold by a correction term that includes the head change rate and the influence of power change. The influence factors of the head change rate and power change are and ;

[0029] The life correction unit, where the temperature threshold is calculated by multiplying the initial temperature threshold by a non-linear correction term that includes the material deterioration coefficient, the ratio of the stress borne by the equipment to the critical stress, the damage index, and the cumulative duration.

[0030] As a preferred solution, the resource matching engine uses an improved multi-objective optimization algorithm to generate a scheduling plan. The objective function is to minimize the weighted sum of the total transportation cost and the longest transportation time, and the constraint condition is that the product of each resource consumption coefficient and the decision variable meets the minimum resource demand threshold.

[0031] As a preferred solution, the visual interaction interface includes:

[0032] The three-dimensional equipment digital twin model: The fault influence intensity is calculated by accumulating the product of the intensity coefficients of each fault source and the Gaussian decay function. The decay coefficient controls the decay speed of the fault influence range with distance.

[0033] The operation confirmation mechanism: The temporary verification code is generated by taking the modulus of the encrypted combination of the Coordinated Universal Time timestamp, the device unique key, and a random number through a hash function to obtain a six-digit number.

[0034] As a preferred solution, the anomaly detection system also includes:

[0035] State assessment report generation module: The remaining service life is calculated by subtracting the cumulative amount of various damages from the design life and then dividing by a correction term that includes an environmental correction factor and a temperature acceleration factor.

[0036] A method for abnormal monitoring of a hydro-generator unit, which uses a system for abnormal monitoring of a hydro-generator unit to perform abnormal monitoring of the hydro-generator unit, including the following steps:

[0037] S1. Data acquisition;

[0038] S2. Data preprocessing and feature extraction;

[0039] S3. Data fusion and analysis;

[0040] S4. Threshold dynamic adjustment and early warning;

[0041] S5. Scheduling decision-making and emergency response;

[0042] S6. Visual display and operation confirmation;

[0043] S7. State assessment report generation.

[0044] It can be seen from the technical solution provided by the present invention described above that the system and method for abnormal monitoring of a hydro-generator unit provided by the present invention have the following beneficial effects:

[0045] Comprehensive and accurate monitoring: By deploying a sensor module covering a vibration sensor array, a temperature sensor group, a pressure sensor, and a flow sensor at key parts of the hydro-generator unit body, and combining with a hydrological monitoring unit at the hydraulic structure, the operation data of the unit and the surrounding hydrological data can be collected in all directions; at the same time, the edge computing node can receive and preprocess these data in real time to ensure the accuracy and timeliness of the data, providing a reliable basis for subsequent analysis;

[0046] Efficient and intelligent early warning: The dynamic early warning model library in the cloud analysis platform uses advanced algorithms, such as a vibration trend prediction sub-model of an improved LSTM, a temperature anomaly classification sub-model driven by a weighted support vector machine, and a fault mode recognition sub-model based on transfer learning, to deeply mine the potential information in the data; the adaptive threshold adjustment module dynamically corrects the alarm threshold according to the equipment operation duration and maintenance records, greatly improving the accuracy and timeliness of early warning and effectively reducing the probability of false alarms and missed alarms;

[0047] Optimize scheduling decisions: The resource matching engine in the scheduling decision center connects to the spare parts inventory database and the personnel qualification database, and uses an improved multi-objective optimization algorithm to generate a scheduling plan, fully balancing cost and timeliness to achieve efficient allocation of resources; The emergency plan generation module can quickly generate a comprehensive emergency plan according to the warning level, including equipment shutdown suggestions, spare parts allocation plans, and emergency repair personnel scheduling instructions, significantly improving the efficiency of fault handling;

[0048] Enhance communication guarantee: The industrial communication network uses the OPCUA protocol to connect each module and is equipped with a time-sensitive network channel, which effectively guarantees the real-time and stable transmission of key data, ensures smooth information interaction between all parts of the system, and enables efficient collaborative work;

[0049] Improve the visual interaction experience: The three-dimensional device twin model of the visual interaction interface uses a physically based rendering algorithm to implement fault propagation simulation, enabling operators to intuitively understand the device status and the development trend of faults; The operation confirmation mechanism uses a two-factor dynamic verification algorithm to generate a temporary verification code, enhancing the security and reliability of operations;

[0050] Realize equipment status evaluation and life prediction: The status evaluation report generation module uses an improved remaining life prediction algorithm to accurately predict the remaining life of the equipment by comprehensively considering various factors, which helps to plan equipment maintenance and updates in advance, reduce the risk of sudden equipment failures, and ensure the long-term stable operation of the hydro-generator set. Brief Description of the Drawings

[0051] Figure 1 It is a schematic diagram of the overall structure of a system and method for abnormal monitoring of a hydro-generator set according to the present invention. Detailed Embodiments

[0052] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0053] To better understand the above technical solutions, the above technical solutions will be described in detail below in conjunction with the drawings of the specification and the specific embodiments.

[0054] As Figure 1 shown, an embodiment of the present invention provides a system for abnormal monitoring of a hydro-generator set, including a sensor module, an edge computing node, a cloud analysis platform, and a scheduling decision center.

[0055] In this embodiment, the sensor module: includes a vibration sensor array, a temperature sensor group, a pressure sensor, and a flow sensor arranged at key parts of the hydro-generator set body, as well as a hydrological monitoring unit arranged in the hydraulic structure;

[0056] Furthermore, the sensor module plays a crucial role in data acquisition in the abnormal monitoring system of the hydro-generator set, and its specific composition is as follows:

[0057] Vibration sensor array:

[0058] Axially arranged acceleration sensor group: Installed at equal intervals along the axis of the main shaft, and the interval satisfies the critical wavelength constraint condition , where, is the interval between adjacent acceleration sensors; is the wavelength corresponding to the lowest characteristic frequency of the hydro-generator set; This arrangement can effectively capture the vibration information in the axial direction of the main shaft, provide key data for subsequent analysis, and ensure the accurate monitoring of the axial vibration state of the unit;

[0059] Radially arranged composite sensor: Integrates vibration and displacement detection functions, and is equipped with a temperature compensation unit; Its integrated functions enable it to simultaneously obtain the vibration and displacement conditions in the radial direction of the unit, while the temperature compensation unit can reduce the interference of temperature changes on the measurement results, improve the accuracy and stability of the measurement, and thus more comprehensively and accurately reflect the radial operating state of the unit;

[0060] Phase synchronization acquisition module: Configured to align the clocks of the vibration signals at each measurement point, and the adaptive adjustment range of the sampling frequency is 1 kHz - 20 kHz;

[0061] The clock alignment ensures the time consistency of the vibration signals at different measurement points, facilitating subsequent data analysis and processing; The adaptively adjusted sampling frequency can flexibly collect data according to the changes in the operating state of the unit, ensuring that vibration signals with appropriate accuracy can be obtained under different working conditions;

[0062] Temperature sensor group: Used to monitor the temperature of key parts of the hydro-generator set, capable of detecting abnormal temperature changes in a timely manner; It is distributed at various key parts of the unit, such as bearings, windings and other places prone to heat generation. By continuously monitoring the temperature data, it provides a basis for judging whether there is an overheating fault in the equipment, ensuring the safe and stable operation of the unit;

[0063] Pressure sensor: Can measure the pressure conditions inside the hydro-generator set, such as oil pressure, water pressure, etc.; Pressure data is crucial for evaluating the operating performance of the unit and detecting potential leakage and other problems, providing data support for timely detection of abnormal pressure changes, and helping to prevent equipment failures caused by pressure problems;

[0064] Flow sensor: Monitor the flow rate of liquids or gases related to the hydro-generator set, such as the cooling water flow rate, gas flow rate, etc.; The flow data can reflect the working efficiency and operating status of the unit. By analyzing the flow rate changes, it is possible to determine whether the cooling system and other components are working properly, ensuring that the unit operates under appropriate flow conditions;

[0065] Hydrological monitoring unit: Installed in hydraulic structures, used to monitor hydrological data related to the operation of the hydro-generator set, such as water level, water flow velocity, water flow rate, etc.; These hydrological data are very important for understanding the working environment and water inlet conditions of the unit, and can provide background information for the operation adjustment and fault analysis of the unit. For example, changes in water level may affect the output and stability of the unit, and the data provided by the hydrological monitoring unit helps to take corresponding measures in advance;

[0066] In summary, the sensor module comprehensively collects the operation data of the hydro-generator set through various types of sensors, providing a rich and accurate data basis for subsequent abnormal monitoring and analysis.

[0067] In this embodiment, the edge computing node: is configured to receive and preprocess sensor data, including a data verification module, a feature extraction module, and a local storage unit;

[0068] Furthermore, the edge computing node plays a crucial role in the middle of the hydro-generator set abnormal monitoring system, mainly responsible for receiving and preprocessing the data collected by the sensors. Its specific composition and functions are as follows:

[0069] Data verification module: The core function of this module is to ensure the accuracy and reliability of the data collected by the sensors; It will use a series of established verification algorithms and rules to strictly screen various types of data transmitted from the sensor module; For example, for the vibration data transmitted by the vibration sensor, it will check whether the numerical range of the data is within a reasonable interval and whether the change trend of the data conforms to the logic of the normal operation of the unit; If abnormal data is found, such as data loss, data values exceeding the normal threshold, or extremely abnormal data fluctuations, this module will mark it in time or take corresponding error correction measures, such as making reasonable estimation and supplementation based on the historical trend of the data and the data of surrounding sensors, so as to provide a high-quality data basis for subsequent data analysis;

[0070] Feature extraction module: Its main function is to extract feature information that is crucial for judging the operating state of the unit from a large amount of original sensor data. For vibration data, it may extract features such as vibration frequency, amplitude, and phase. For temperature data, it will extract features such as the mean value, peak value, and change rate of temperature. By adopting advanced signal processing technologies and data analysis algorithms, such as wavelet transform and Fourier transform, complex original data can be transformed into more representative and analytically valuable feature vectors. These extracted features not only significantly reduce the amount of data and the burden of subsequent data transmission and processing, but also can more intuitively reflect the operating state of the unit, providing strong support for subsequent anomaly monitoring and fault diagnosis.

[0071] Local storage unit: It has the function of storing the data that has been verified and partially processed, as well as the extracted feature information. In the case of network failures or data transmission delays, the local storage unit can serve as a temporary repository for data to ensure that the data is not lost. At the same time, it can also store some historical data, which is of great value for analyzing the operating trend of the unit, conducting long-term performance evaluations, and comparing current data with historical normal data to determine whether there are anomalies. For example, by comparing the current vibration characteristics with the vibration characteristics under the same working conditions in the past period, potential abnormal changes can be more accurately detected, providing a basis for taking timely maintenance measures.

[0072] In summary, through the collaborative work of its internal modules, the edge computing node effectively preprocesses and preliminarily analyzes the sensor data, improves the quality and usability of the data, and lays a solid foundation for the efficient operation of the entire anomaly monitoring system.

[0073] In this embodiment, the cloud analysis platform is provided with a multi-source data fusion module, a dynamic early warning model library, and an adaptive threshold adjustment module.

[0074] Among them, the multi-source data fusion module is used to integrate the unit operation data, hydrological data, and power grid load data.

[0075] Furthermore, the multi-source data fusion module plays a key role in the core integration of the water turbine generator unit anomaly monitoring system, and its specific functions and implementation methods are as follows:

[0076] Data integration function:

[0077] This module is capable of receiving data from different data sources and conducting unified aggregation and collation. This includes unit operation data, such as parameters like the rotational speed, output power, current, and voltage of the generator, which reflect the electrical and mechanical operating states of the unit itself; hydrological data, such as water level height, water flow velocity, water flow rate, and water quality conditions, which reflect the impact of the external water conservancy environment where the unit is located on its operation; and grid load data, such as the real-time load demand and power factor of the grid, because changes in grid load have an important impact on the operating conditions of the hydro-generator set. By collecting and integrating these multiple types of data, it provides a comprehensive data foundation for subsequent comprehensive analysis.

[0078] Data preprocessing and standardization:

[0079] Before data fusion, the multi-source data fusion module performs preprocessing operations on data from different sources. Since the data formats, precisions, and dimensions collected by different sensors and data sources may vary, data cleaning is required to remove noise data and outliers to improve data quality. For example, some electrical data points that deviate significantly from the normal range due to electromagnetic interference are removed. At the same time, data standardization processing is carried out to convert data with different dimensions into a unified standard, enabling various types of data to be analyzed and compared on the same scale. For example, temperature data is converted from Celsius to the Kelvin scale and its numerical range is normalized to the [0,1] interval, so that the relationship between data can be more accurately reflected in subsequent analysis models.

[0080] Data association and fusion algorithms:

[0081] This module uses advanced algorithms to achieve in-depth data fusion. For example, the method based on Bayesian network can establish a probability relationship model between different data, and can infer the states of other related data based on known partial data, thus achieving a more comprehensive assessment of the unit's operating state. Another example is the use of Kalman filtering technology to fuse and estimate the dynamic operation data of the unit. Based on considering measurement noise and system dynamic characteristics, the estimated value of the unit's state is continuously updated and optimized to improve the accuracy and real-time performance of monitoring. Through these data association and fusion algorithms, the originally scattered multi-source data is integrated into an organic whole, providing more accurate and comprehensive information support for subsequent anomaly monitoring and fault diagnosis, and effectively enhancing the perception and judgment ability of the entire monitoring system for the operating state of the hydro-generator set.

[0082] Furthermore, the dynamic warning model library includes a vibration trend prediction sub-model based on improved LSTM, a temperature anomaly classification sub-model driven by weighted support vector machines, and a fault mode recognition sub-model based on transfer learning.

[0083] Specifically, the dynamic early warning model library is a key part of the abnormal monitoring system for hydro-generator units. Its working process covers multiple important links, and each sub-model works together as follows:

[0084] Establishing the equipment health baseline: The improved hybrid wavelet packet decomposition algorithm is used to extract features. In this algorithm, , where is the wavelet coefficient of the -th layer and the -th node in the improved hybrid wavelet packet decomposition algorithm; is the input discrete time series signal; is the complex conjugate of the modified Morlet wavelet basis function at the -th layer and the -th node; is the attenuation factor, which is used to adjust the attenuation speed of the exponential decay term; is the serial number of the discrete time series; is the discrete time serial number corresponding to the center position of the fault sensitive area. In this way, features can be accurately extracted from complex operation data, laying a foundation for subsequent analysis, and establishing a health baseline that can reflect the normal operation state of the equipment, which is used as an important reference standard for judging whether the equipment is abnormal;

[0085] Real-time abnormal detection: The weighted Mahalanobis distance is used to calculate the feature deviation degree, and the formula is , where is the weighted Mahalanobis distance, which is used to measure the deviation degree of the real-time feature vector relative to the benchmark; is the real-time feature vector; is the benchmark mean vector, representing the mean of the feature vector under normal conditions; is the transpose of the vector ; is the covariance matrix 's inverse matrix; is the time-varying weight coefficient, which is used to strengthen the influence of recent data, and its value changes with time , which can strengthen the influence of recent data. By continuously calculating the feature deviation degree, the deviation between the unit operation data and the normal state can be found in time, so as to effectively capture the occurrence of abnormal situations;

[0086] Multi-model collaborative decision-making: When the alarm confidence level of the sub-model exceeds the threshold, the cross-validation process is started. The final alarm probability calculation formula is

[0087] , where is the final alarm probability; is the dynamic weight coefficient, which is used to balance the weights after combining the vibration trend prediction sub-model based on the improved LSTM and the other two sub-models; is the alarm probability output by the vibration trend prediction sub-model based on the improved LSTM; is the alarm probability output by the temperature anomaly classification sub-model driven by the weighted support vector machine; is the alarm probability output by the fault mode recognition sub-model based on transfer learning; is the activation function, which maps the input value to interval; Through the collaborative work and cross-validation of multiple models, considering various factors comprehensively, the accuracy and reliability of the alarm are improved, the false alarm and missed alarm situations are reduced, and it is ensured that the abnormal state of the hydro-generator unit can be warned in a timely and accurate manner;

[0088] The dynamic early warning model library plays an important role in the abnormal monitoring of hydro-generator units through its unique algorithm and multi-model collaborative mechanism, providing a strong guarantee for the safe and stable operation of the equipment;

[0089] Furthermore, the adaptive threshold adjustment module dynamically corrects the alarm threshold according to the equipment operation duration and maintenance records;

[0090] Specifically, the adaptive threshold adjustment module is crucial in the abnormal monitoring system of hydro-generator units. It can dynamically correct the alarm threshold according to the actual operation situation of the equipment and is mainly composed of a working condition adaptation unit and a life correction unit:

[0091] Working condition adaptation unit:

[0092] Adopts a vibration threshold correction formula with multi-parameter coupling , where is the vibration threshold after working condition correction; is the reference vibration threshold, which is the vibration threshold set under standard working conditions; is the working condition influence factor of the head change rate on the vibration threshold; is the working condition influence factor of the power change on the vibration threshold; is the head change rate; is the real-time power of the hydro-generator unit; is the rated power of the hydro-generator unit;

[0093] When the head change rate or power changes during the operation of the unit, this unit will adjust the vibration threshold in real time according to the above formula; For example, when the head rises rapidly or the power fluctuates greatly, through the action of the corresponding working condition influence factor, the vibration threshold is appropriately increased to avoid false alarms caused by normal vibration fluctuations due to working condition changes, ensuring the accuracy and rationality of the alarm, and enabling the monitoring system to better adapt to the operation characteristics of the unit under different working conditions;

[0094] Lifetime correction unit:

[0095] Adjust the temperature threshold based on the non - linear cumulative damage model. The formula is:

[0096] , where is the temperature threshold after lifetime correction; is the initial temperature threshold; is the material deterioration coefficient; is the equivalent stress endured by the equipment in the th time period; is the critical stress of the equipment material; is the damage index, used to describe the non - linear relationship between stress and damage; is the duration of the th time period;

[0097] As the operation time of the equipment increases and the material gradually deteriorates, the upper limit of the temperature that the equipment can withstand will change. This unit continuously tracks the stress conditions endured by the equipment in different time periods, calculates the cumulative damage amount using the non - linear cumulative damage model, and then dynamically adjusts the temperature threshold. For example, when the equipment operates in a high - stress environment for a long time, the material deterioration accelerates, and the temperature threshold will be correspondingly reduced, so as to more sensitively monitor the potential equipment failure hazards caused by temperature rise and ensure the safe operation of the equipment.

[0098] In summary, through the coordinated action of the working condition adaptation unit and the lifetime correction unit, the adaptive threshold adjustment module can flexibly adjust the alarm threshold according to the operating conditions and equipment lifetime status of the hydro - generator set, effectively improving the reliability and adaptability of the abnormal monitoring system.

[0099] In this embodiment, the dispatching decision - making center includes a resource matching engine, an emergency plan generation module, a visual interaction interface, and an industrial communication network;

[0100] Among them, the resource matching engine is connected to the spare part inventory database and the personnel qualification database;

[0101] Specifically, the resource matching engine plays a key role in the dispatching decision - making center of the hydro - generator set abnormal monitoring system. Its core function is to generate an efficient and reasonable dispatching plan, mainly achieved through specific objective functions and constraint conditions:

[0102] Objective function: Adopt an improved multi - objective optimization algorithm, and its objective function is , where this formula is the objective function, and the goal is to minimize its value; is the The transportation cost of a transportation task; is a decision variable, taking values of 0 or 1, where 1 indicates selecting the transportation task and 0 indicates not selecting it; is the time penalty factor, used to balance transportation cost and transportation time; For the transportation time of the th transportation route; is the total number of transportation tasks;

[0103] Constraint conditions: , where represents the constraint condition; is the resource constraint coefficient, indicating the consumption coefficient of the rd transportation task for the th type of resource; is a decision variable; For the th lowest demand threshold of the resource; is the number of transportation tasks; is the number of resource types; These constraint conditions ensure the feasibility and rationality of the generated scheduling plan in resource allocation, avoid over-allocation or shortage of resources, ensure that spare parts and personnel can be allocated in a timely and reasonable manner when the water turbine generator set has an abnormality and needs emergency repair, improve the emergency repair efficiency, and reduce the downtime;

[0104] In summary, through the carefully designed objective function and constraint conditions, the resource matching engine uses an improved multi-objective optimization algorithm to provide scientific and effective decision support for resource scheduling in the case of abnormalities of the water turbine generator set, and plays an important role in ensuring the stable operation of the unit;

[0105] Furthermore, the emergency plan generation module is configured to generate an emergency plan including equipment shutdown suggestions, spare part allocation plans, and emergency repair personnel scheduling instructions according to the warning level;

[0106] Specifically: The emergency plan generation module plays a key role in the scheduling decision center of the water turbine generator set abnormality monitoring system. Its core function is to quickly and reasonably generate a comprehensive emergency plan according to the warning level, as follows:

[0107] Equipment shutdown suggestion generation: When the system receives the warning information from the dynamic warning model library, the emergency plan generation module will judge whether it is necessary to suggest equipment shutdown according to the warning level, and determine the timing and method of shutdown; for high-level warnings, it indicates that there may be serious potential faults in the unit. At this time, the module will immediately generate a suggestion for emergency shutdown of the equipment to avoid more serious damage to the unit caused by the further deterioration of the fault; when determining the shutdown method, the operating status of the unit and the actual on-site situation will be considered. For example, if the unit is in a full-load operating state, the shutdown process may require gradually unloading the load before shutting down to ensure the safety and stability of the shutdown process; at the same time, the shutdown sequence of some key equipment will also be reasonably arranged to prevent secondary faults caused by improper shutdown sequence;

[0108] Spare part allocation plan formulation: This module works closely with the resource matching engine. According to the possible fault types and damaged parts of the equipment prompted by the warning information, combined with the information in the spare part inventory database, a detailed spare part allocation plan will be generated; first, the types and quantities of the required spare parts will be accurately determined; for some vulnerable spare parts, such as bearings, seals, etc., they will be estimated according to historical fault data and the current operating condition of the equipment; then, considering the storage location and transportation route of the spare parts, using the optimization algorithm of the resource matching engine, a plan with low transportation cost and short transportation time will be selected to ensure that the spare parts can be delivered to the site in time; during the transportation process, the transportation status of the spare parts will also be tracked to adjust the transportation strategy in time to ensure that the spare parts arrive on time;

[0109] Emergency repair personnel dispatch instruction generation: The emergency plan generation module will screen out emergency repair personnel with corresponding skills and experience from the personnel qualification database according to the warning level and fault type, and generate detailed dispatch instructions; for complex electrical faults, professional electrical engineers will be dispatched; for mechanical faults, mechanical maintenance experts will be arranged; at the same time, according to the urgency and difficulty of the fault, the number of personnel will be reasonably allocated and an emergency repair team will be formed; during the dispatch process, the responsibilities and task assignments of each emergency repair personnel will be clarified to ensure the efficient and orderly development of the emergency repair work; for example, when repairing the fault of the large turbine blade, an experienced mechanical engineer will be arranged to be responsible for on-site command, the welder will be responsible for blade repair work, and the auxiliary personnel will be responsible for tool preparation and on-site cleaning, etc. The repair efficiency will be improved through clear division of labor;

[0110] In summary, the emergency plan generation module generates a comprehensive and targeted emergency plan by comprehensively considering various factors such as warning level, equipment status, spare part inventory and personnel qualification, providing a strong guarantee for the handling of abnormal situations of the hydro-generator set, effectively reducing the losses caused by faults, and improving the reliability and availability of the unit;

[0111] Furthermore, the visual interaction interface provides three-dimensional equipment status display and operation confirmation functions;

[0112] Specifically, the visual interaction interface plays an important role in the abnormal monitoring system of the water turbine generator set, providing users with an intuitive display of the equipment status and a secure operation confirmation function. Its main components and functions are as follows:

[0113] Three-dimensional equipment twin model:

[0114] The physical-based rendering algorithm is used to implement the fault diffusion simulation, and its formula is

[0115] , where is the fault influence intensity at the coordinate in the three-dimensional equipment twin model; is the total number of fault sources, covering various factors that may cause unit failures, such as mechanical component wear, electrical fault points, etc.; is the intensity coefficient of the th fault source, reflecting the initial severity of this fault source. For example, the intensity coefficient of a serious bearing damage fault source will be relatively high; is the position coordinate of the th fault source; is the influence range diffusion coefficient of the th fault source, which controls the attenuation speed of the fault influence range with distance and can intuitively display the propagation trend of the fault within the unit. Through this model, users can clearly see the overall structure of the equipment, as well as the possible fault locations, influence ranges, and intensities in a three-dimensional visualization manner, facilitating a quick understanding of the equipment's health status and potential risks;

[0116] Operation confirmation mechanism:

[0117] The dual-factor dynamic verification algorithm is used to generate a temporary verification code, and the formula is

[0118] , where is the generated temporary verification code, which is a six-digit number; is the hash function, which uses its powerful encryption function to encrypt the input information to ensure the security of the information; is the Coordinated Universal Time timestamp, which provides the current accurate time and serves as a dynamic factor for verification, increasing the timeliness and randomness of the verification code; is the bitwise exclusive OR operator; is the unique key of the device, ensuring the relevance of the verification code to a specific device; is a random number generated based on the random seed , further enhancing the randomness of the verification code; represents taking the modulus of the result , the result is limited within a six-digit range, which is convenient for users to input and verify; when users perform critical operations such as remotely controlling the equipment to stop or modifying important parameters, the system will generate a temporary verification code, and users can only execute the operation by inputting the correct verification code, effectively preventing misoperations and illegal operations, and ensuring the safe and stable operation of the equipment;

[0119] In summary, through the three-dimensional device twin model and the operation confirmation mechanism, the visual interaction interface provides users with rich information display and secure operation guarantee, greatly improving the usability and security of the hydro-generator unit abnormal monitoring system;

[0120] Furthermore, the industrial communication network uses the OPC UA protocol to connect each module and is equipped with a time-sensitive network channel to ensure the transmission of critical data;

[0121] Specifically, the industrial communication network is a key part to achieve efficient and stable data transmission among modules in the hydro-generator unit abnormal monitoring system, and its main features and functions are as follows:

[0122] Protocol selection: Use the OPC UA protocol to connect each module; OPC UA (Open Platform Communications Unified Architecture) is an advanced industrial communication protocol with advantages such as cross-platform, high reliability, and good interoperability; it can achieve seamless communication between devices and systems of different manufacturers, ensuring accurate data transmission among various modules such as sensor modules, edge computing nodes, cloud analysis platforms, dispatching decision-making centers, and visual interaction interfaces; for example, when transmitting data such as temperature, pressure, and vibration collected by sensors to the edge computing node for preprocessing, the OPC UA protocol can ensure the integrity and timeliness of the data, avoiding data loss or transmission errors;

[0123] Guarantee of time-sensitive network channel: There is a time-sensitive network (TSN) channel to ensure the transmission of critical data; The operation status monitoring of the hydro-generator unit requires real-time and reliable transmission of some critical data, such as high-frequency vibration data collected by vibration sensors and abnormal data that may indicate an impending fault; The TSN channel ensures that these critical data can be accurately transmitted to the corresponding module for analysis and processing within the specified time by providing deterministic network latency and low jitter characteristics; for example, when the unit has a sudden vibration anomaly, the vibration data is quickly transmitted to the dynamic early warning model library of the cloud analysis platform via the TSN channel, so as to trigger the early warning mechanism in time and gain valuable time for the safe operation of the unit; At the same time, the TSN channel can still ensure the priority transmission of critical data under high network load, effectively avoiding data transmission delays or losses caused by network congestion, and improving the reliability and response speed of the entire monitoring system;

[0124] In summary, the industrial communication network, relying on the generality of the OPC UA protocol and the ability of the TSN channel to guarantee key data, provides a solid communication foundation for the collaborative work among the modules of the hydro-generator unit abnormal monitoring system, ensuring that the system can operate stably and efficiently, and realizing the real-time monitoring and abnormal early warning of the unit operation status.

[0125] In this embodiment, the system further includes:

[0126] A status evaluation report generation module, which plays an important role in the hydro-generator unit abnormal monitoring system. It mainly uses an improved remaining life prediction algorithm to evaluate the unit status and generate a report. The status evaluation report generation module adopts the improved remaining life prediction algorithm:

[0127] , where, is the remaining service life of the hydro-generator unit, which is a key indicator for evaluating the health status and operation reliability of the unit; is the design life of the hydro-generator unit, that is, the total theoretical operating duration of the equipment, which provides a reference value for the calculation of the remaining life; is the cumulative amount of the type of damage, covering various damage factors that may affect the unit life, such as mechanical wear, electrical aging, material fatigue, etc. By accurately calculating and accumulating these damage amounts, the loss degree of the unit during actual operation can be reflected; is the total number of damage types, indicating that multiple different types of damage situations need to be comprehensively considered; is the environmental correction coefficient, which is used to consider the influence of environmental factors on the equipment life. For example, factors such as the temperature, humidity, and water quality of the environment where the unit is located will accelerate or slow down the aging and damage process of the equipment. Through this coefficient, the equipment life under different environments can be reasonably corrected; is the temperature acceleration factor, which reflects the influence degree of temperature on the equipment aging speed. Since temperature is one of the important factors affecting the equipment life, especially in a high-temperature environment, the aging speed of the equipment will be significantly accelerated. This factor can highlight the key role of temperature in life prediction; is the average operating temperature of the hydro-generator unit, which provides the actual data basis for the calculation of the temperature acceleration factor;

[0128] Report generation process: The status assessment report generation module first collects data from various sources such as the sensor module, edge computing nodes, and cloud analysis platforms, including the operating parameters of the unit, historical fault records, environmental monitoring data, etc.; then, based on the above algorithms, it deeply analyzes and calculates these data to obtain the remaining service life of the unit and the current health status assessment results; in the generated report, in addition to clearly giving the value of the remaining life, it will also detail the key factors affecting the life, such as which components have more serious damage, the potential risks of the current operating environment to the equipment, etc.; at the same time, it will provide maintenance suggestions based on the assessment results, such as suggesting to replace key components in advance or conduct a comprehensive overhaul of the equipment when the remaining life reaches a certain threshold, and proposing improvement measures for the current operating environment, such as strengthening ventilation and heat dissipation to reduce the impact of temperature on the equipment, etc., providing comprehensive and scientific decision-making basis for equipment maintenance personnel, helping to reasonably arrange the equipment maintenance plan, improve the reliability and availability of the equipment, and reduce the losses caused by equipment failures;

[0129] In summary, the status assessment report generation module can provide important reference information for the operation and maintenance of hydro-generator units through advanced algorithms and comprehensive data processing, and plays an indispensable role in ensuring the safe and stable operation of the equipment.

[0130] A method for abnormal monitoring of hydro-generator units, which uses a specific system to achieve all-round monitoring of the unit. The specific steps are as follows:

[0131] S1. Data collection:

[0132] With the help of the sensor module, various operating data of the hydro-generator unit are comprehensively collected; among them, the vibration sensor array is reasonably arranged axially and radially to accurately collect the vibration information of the unit; the axial acceleration sensor group is installed at equal intervals along the axis of the main shaft, and the interval satisfies the critical wavelength constraint condition to ensure that the axial vibration characteristics can be effectively captured; the radial composite sensor integrates vibration and displacement detection functions and has a temperature compensation unit, which can accurately obtain the relevant radial information. The phase synchronization acquisition module aligns the clock of the vibration signals at each measurement point, and the sampling frequency is adaptively adjusted between 1 kHz and 20 kHz; at the same time, the temperature sensor group, pressure sensor, and flow sensor respectively monitor the temperature of key parts, the internal pressure, and the flow of relevant liquids or gases, and the hydrological monitoring unit of the hydraulic structure is responsible for collecting hydrological data such as water level and water flow velocity; these data provide rich original materials for subsequent analysis;

[0133] S2. Data preprocessing and feature extraction:

[0134] After the edge computing node receives the sensor data, the data verification module strictly verifies the data using specific algorithms and rules, eliminates outliers and incorrect data, and ensures the accuracy of the data; the feature extraction module adopts advanced signal processing and data analysis algorithms, such as extracting features such as frequency and amplitude from vibration data, and extracting features such as mean value and change rate from temperature data, converting the original data into more representative feature vectors, which not only reduces the data volume but also highlights the key information, laying a foundation for subsequent analysis and judgment;

[0135] S3. Data Fusion and Analysis:

[0136] The multi-source data fusion module of the cloud analysis platform integrates the unit operation data, hydrological data, and power grid load data, performs data cleaning and standardization processing before fusion to make the data from different sources comparable on the same scale; then uses advanced algorithms such as Bayesian network and Kalman filter to achieve deep data fusion, comprehensively analyzes the correlation and influence between various data, and understands the unit operation status more comprehensively; the dynamic early warning model library uses an improved hybrid wavelet packet decomposition algorithm to establish a device health baseline, real-time detects the feature deviation degree through the weighted Mahalanobis distance, and when the alarm confidence level of the sub-model exceeds the threshold, uses the multi-model collaborative decision-making process to calculate the final alarm probability to timely detect abnormal situations;

[0137] S4. Threshold Dynamic Adjustment and Early Warning:

[0138] The adaptive threshold adjustment module dynamically corrects the alarm threshold according to the device operation duration and maintenance records; the working condition adaptation unit adjusts the vibration threshold according to the vibration threshold correction formula coupled with multiple parameters, considering factors such as the head change rate and power change; the life correction unit adjusts the temperature threshold based on the non-linear cumulative damage model, combined with the material deterioration coefficient and the stress borne by the device, etc., to ensure the accuracy and reliability of the alarm, avoid false alarms and missed alarms, and timely send early warning information to relevant personnel;

[0139] S5. Scheduling Decision and Emergency Response:

[0140] After receiving the early warning, the resource matching engine of the scheduling decision center uses an improved multi-objective optimization algorithm, comprehensively considers the transportation cost and time factors, and based on the spare parts inventory and personnel qualification database, generates a reasonable scheduling plan, determines the spare parts transportation task and the deployment of repair personnel; the emergency plan generation module formulates an emergency plan including equipment shutdown suggestions, spare parts allocation plans, and repair personnel scheduling instructions according to the early warning level, guiding the on-site personnel to take actions quickly to ensure the safety of the unit;

[0141] S6. Visualization Display and Operation Confirmation:

[0142] The visual interaction interface uses a three-dimensional device twin model and a physically based rendering algorithm to simulate the fault diffusion process, intuitively displaying the device status and the impact of faults; the operation confirmation mechanism uses a two-factor dynamic verification algorithm to generate a temporary verification code for identity verification when users perform critical operations, preventing misoperations and illegal operations and ensuring the safe and stable operation of the system;

[0143] S7. Generation of the status evaluation report:

[0144] The status evaluation report generation module uses an improved remaining life prediction algorithm, comprehensively considering the device design life, various damage accumulations, environmental factors, temperature effects, etc., calculates the remaining service life of the hydro-generating unit, and generates a detailed report, providing a scientific basis for device maintenance and management and facilitating the reasonable arrangement of maintenance plans and resource allocation;

[0145] Through the above series of steps, this abnormal monitoring method can achieve efficient and accurate monitoring of the hydro-generating unit, timely detect and handle abnormal situations, and ensure the safe and stable operation of the unit.

[0146] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A system for abnormal monitoring of a hydro-generating unit, characterized in that: Including: Sensor module: It includes a vibration sensor array, a temperature sensor group, a pressure sensor, and a flow sensor arranged at key parts of the water turbine generator set body, as well as a hydrological monitoring unit set in the hydraulic structure; Edge computing node: Configured to receive and preprocess sensor data, including a data verification module, a feature extraction module, and a local storage unit; Cloud analysis platform: Provided with: Multi-source data fusion module: Used to integrate unit operation data, hydrological data, and grid load data; Dynamic warning model library: Includes a vibration trend prediction sub-model based on improved LSTM, a temperature anomaly classification sub-model driven by a weighted support vector machine, and a fault mode recognition sub-model based on transfer learning; The working method of the dynamic warning model library includes: Establishing an equipment health baseline: Using an improved hybrid wavelet packet decomposition algorithm to extract features; Real-time anomaly detection: Measuring the deviation degree of the real-time feature vector relative to the benchmark through the weighted Mahalanobis distance; Multi-model collaborative decision-making: When the alarm confidence level of the sub-model exceeds the threshold, the calculation formula for the final alarm probability is: , where is the final alarm probability; is the dynamic weight coefficient; is the alarm probability output by the vibration trend prediction sub-model based on the improved LSTM; is the alarm probability output by the temperature anomaly classification sub-model driven by the weighted support vector machine; is the alarm probability output by the fault mode recognition sub-model based on transfer learning; is the activation function that maps the input value to interval; Adaptive threshold adjustment module: Dynamically corrects the alarm threshold according to the equipment operation duration and maintenance records. The adaptive threshold adjustment module includes: Working condition adaptation unit, the vibration threshold is calculated by multiplying the benchmark vibration threshold by a correction term including the head change rate and the power change impact; Lifetime correction unit, the temperature threshold is calculated by multiplying the initial temperature threshold by a non-linear correction term including the material degradation coefficient, the ratio of the stress borne by the equipment to the critical stress, the damage index, and the duration accumulation; Dispatch decision center: Including: Resource matching engine: Connects to the spare parts inventory database and the personnel qualification database; Emergency plan generation module: Configured to generate an emergency plan including equipment shutdown suggestions, spare parts allocation plans, and emergency repair personnel dispatch instructions according to the warning level; Visualization interaction interface: Provides three-dimensional equipment status display and operation confirmation functions; Industrial communication network: Connects each module using the OPCUA protocol and is provided with a time-sensitive network channel to ensure the transmission of key data.

2. The system for abnormal monitoring of a hydro-generating unit according to claim 1, wherein: The vibration sensor array adopts a spatially distributed layout scheme, including: Axially arranged acceleration sensor group: Installed at equal intervals along the main shaft axis, and the distance d between adjacent acceleration sensors is less than one-fourth of the wavelength corresponding to the lowest characteristic frequency of the water turbine generator set; Radially arranged composite sensor: Integrates vibration and displacement detection functions and is provided with a temperature compensation unit; Phase synchronization acquisition module: Configured to align the clock of the vibration signals at each measurement point, and the sampling frequency adaptive adjustment range is 1kHz - 20kHz.

3. The system for abnormal monitoring of a hydro-generating unit according to claim 1, wherein: The resource matching engine uses an improved multi-objective optimization algorithm to generate a scheduling plan, and the objective function is to minimize the weighted sum of the total transportation cost and the longest transportation time, and the constraint condition is that the product of each resource consumption coefficient and the decision variable meets the minimum resource demand threshold.

4. A system for abnormal monitoring of a hydro-generating unit according to claim 1, characterized in that: The visualization interaction interface includes: Three-dimensional equipment twin model: The fault impact intensity is calculated by the cumulative product of the intensity coefficients of each fault source and the Gaussian attenuation function, and the attenuation coefficient controls the attenuation speed of the fault impact range with distance; Operation confirmation mechanism: The temporary verification code is generated by taking the modulus of the encryption of the combination of Coordinated Universal Time timestamp, device unique key, and random number through a hash function to obtain a six-digit number.

5. The system for abnormal monitoring of a hydro-generating unit according to claim 1, wherein: It also includes: State evaluation report generation module: The remaining service life is calculated by subtracting the cumulative amount of various damages from the design life and dividing by the correction term including the environmental correction factor and the temperature acceleration factor.

6. A method for abnormal monitoring of a hydro-generating unit, characterized in that: The method for abnormal monitoring of the hydro-generator set uses the system for abnormal monitoring of the hydro-generator set described in any one of claims 1-5 to perform abnormal monitoring of the hydro-generator set, including the following steps: S1. Data acquisition; S2. Data preprocessing and feature extraction; S3. Data fusion and analysis; S4. Threshold dynamic adjustment and early warning; S5. Scheduling decision-making and emergency response; S6. Visual display and operation confirmation; S7. State evaluation report generation.

Images