Water turbine runner real-time monitoring and intelligent operation and maintenance method and system based on knowledge graph

Abstract

The invention discloses a water turbine runner real-time monitoring and intelligent operation and maintenance method and system based on a knowledge graph, and belongs to the technical field of hydroelectric power generation equipment state monitoring. The problems of interpretable monitoring and self-adaptive maintenance decision generation of the whole life cycle state of the turbine runner are solved. A pressure sensor, a vibration sensor array and an acoustic emission sensor are arranged on a turbine runner; the signal acquisition module acquires running state parameters of the turbine runner in real time to obtain original data; after the graph construction module preprocesses the original data, an equipment body model of the water turbine runner is established, and a knowledge graph of the water turbine runner is generated; the monitoring analysis module analyzes the state of the turbine runner in real time to obtain the output of the monitoring analysis module; and the intelligent operation and maintenance module generates maintenance strategy suggestions through similarity calculation matching historical cases according to the output of the monitoring analysis module. According to the method, the technical span of monitoring the state of the turbine runner from traditional threshold judgment to cognitive intelligent decision is realized.

Description

Technical Field

[0001] This invention belongs to the field of hydropower equipment condition monitoring technology, specifically relating to a method and system for real-time monitoring and intelligent operation and maintenance of turbine runners based on knowledge graphs. Background Technology

[0002] As a core component of hydroelectric generator sets, the turbine runner's operating status directly impacts unit efficiency and equipment safety. Traditional monitoring methods primarily rely on single physical quantity measuring devices such as vibration sensors and pressure transmitters, using threshold alarms for fault warnings. These technologies suffer from limitations such as limited monitoring dimensions and lack of data correlation, making it difficult to effectively reflect the complex operating conditions of the runner components. Existing intelligent operation and maintenance systems often employ statistical analysis models based on historical data, lacking a deep integration of equipment mechanistic knowledge and operational data, resulting in insufficient interpretability of condition assessment results.

[0003] The current field of turbine runner monitoring faces the following technical bottlenecks: First, the integration efficiency of multi-source heterogeneous data is low, and physical quantities such as vibration signals, cavitation noise, stress, and strain collected by sensors lack a unified characterization framework in terms of temporal synchronization and spatial correlation. Second, expert experience and knowledge are difficult to effectively structure, and equipment maintenance decisions rely excessively on manual experience judgment, leading to a risk of knowledge gaps in transmission. Third, existing monitoring systems have limited processing capabilities for real-time data streams, failing to achieve rapid capture and inference of millisecond-level anomalies. Furthermore, traditional methods have not established component-level state evolution maps, making it difficult to support dynamic tracking of the entire lifecycle health status. Summary of the Invention

[0004] The problem to be solved by this invention is to realize interpretable monitoring and adaptive maintenance decision generation of the entire life cycle status of a turbine runner. It proposes a method and system for real-time monitoring and intelligent operation and maintenance of a turbine runner based on knowledge graph.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A knowledge graph-based method for real-time monitoring and intelligent operation and maintenance of turbine runners includes the following steps:

[0007] S1. Pressure sensors, vibration sensor arrays, and acoustic emission sensors are installed on the turbine runner;

[0008] S2. The signal acquisition module acquires the operating status parameters of the turbine runner in real time to obtain raw data;

[0009] S3. After preprocessing the raw data obtained in step S2, the graph construction module establishes a device model of the turbine runner and generates a knowledge graph of the turbine runner.

[0010] S4. Based on the knowledge graph of the turbine runner obtained in step S2, the monitoring and analysis module analyzes the state of the turbine runner in real time and obtains the output of the monitoring and analysis module;

[0011] S5. The intelligent operation and maintenance module generates maintenance strategy suggestions by matching historical cases through similarity calculation based on the output of the monitoring and analysis module.

[0012] Furthermore, the specific implementation method of step S1 includes the following steps:

[0013] S1.1. The vibration sensor array is arranged at equal angles at the blade roots and hub connection of the turbine runner, with the installation interval angles meeting the following requirements:

[0014]

[0015] in, This refers to the number of blades on the turbine runner. This represents the difference in installation angle between adjacent sensors;

[0016] S1.2. Pressure sensors are installed in the upper crown and lower circulation channel areas of the turbine runner, with a sampling frequency of... Set to 8-10 times the characteristic frequency of the turbine runner, satisfying:

[0017]

[0018] in, This refers to the number of blades on the turbine runner. The frequency corresponding to the rated speed;

[0019] S1.3. Acoustic emission sensors are placed in the following key areas of the turbine runner: the rounded corner area between the blade and the hub; the area near the middle of the blade near the inlet edge; the connection between the blade trailing edge and the cover plate; and the area near the connecting flange between the inner side of the hub and the main shaft.

[0020] Furthermore, the specific implementation method of step S3 includes the following steps:

[0021] S3.1. Establish the equipment body model of the turbine runner. The mathematical expression is:

[0022]

[0023] in, , This represents a set of entities, including equipment entities, state entities, and process entities of the turbine runner. For the m-th specific entity; , Represents a set of relationships between entities; , For a collection of attributes, This is the kth specific attribute of the entity;

[0024] S3.2. An improved TransH algorithm is used to complete the knowledge of the turbine runner's equipment model, generating a knowledge graph of the turbine runner, expressed as:

[0025]

[0026] in, This is the overall vectorized representation of the knowledge graph of a water turbine runner. These represent the head entity vector and the tail entity vector, respectively. Let be the normal vector of the relation hyperplane. A translation vector specific to the relation.

[0027] Furthermore, the specific implementation method of step S4 includes the following steps:

[0028] S4.1. Real-time monitoring data By embedding functions Incremental update of the overall vector of the knowledge graph transformed into that of a water turbine runner;

[0029] S4.2. Design a dynamic reasoning algorithm based on the matching degree of the overall vector of the knowledge graph of the turbine runner. Anomaly detection is performed, and the state of the turbine runner is analyzed in real time. The expression is:

[0030]

[0031] in, This is the current atlas fragment. For reference standard atlas.

[0032] Furthermore, in step S5, the method for generating maintenance strategy suggestions by matching historical cases through similarity calculation involves constructing a case reasoning engine, which uses similarity... The expression is used to calculate the matching historical case library:

[0033]

[0034] in, For the feature vector of the current problem, This is the feature vector of the i-th fault or operating condition case in the historical case library;

[0035] Design a strategy optimization loop based on the feedback matrix. The entity weight distribution of the knowledge graph is dynamically adjusted, expressed as:

[0036]

[0037] in, This represents the weight update value between the i-th entity and the j-th relation in the knowledge graph within the feedback matrix, used for reinforcement learning-based knowledge optimization mechanisms.

[0038] A knowledge graph-based real-time monitoring and intelligent operation and maintenance system for turbine runners, implemented based on the aforementioned knowledge graph-based real-time monitoring and intelligent operation and maintenance method for turbine runners, includes a signal acquisition module, a graph construction module, a monitoring and analysis module, and an intelligent operation and maintenance module, which are connected sequentially.

[0039] The beneficial effects of this invention are:

[0040] The present invention discloses a real-time monitoring and intelligent operation and maintenance method for turbine runners based on knowledge graphs. This method improves the accuracy of fault mode identification by leveraging the semantic reasoning capabilities of knowledge graphs; reduces the false alarm rate by utilizing a dynamic graph update mechanism; and reduces maintenance costs by combining intelligent operation and maintenance strategy optimization. The present invention achieves a technological leap from traditional threshold judgment to cognitive intelligent decision-making in turbine runner status monitoring. Attached Figure Description

[0041] Figure 1 This is a flowchart of a knowledge graph-based real-time monitoring and intelligent operation and maintenance method for turbine runners according to the present invention. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention; that is, the described specific embodiments are merely a part of the embodiments of the invention, and not all of them. The components of the specific embodiments of the invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations, and the invention may also have other embodiments.

[0043] Therefore, the following detailed description of specific embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected specific embodiments of the invention. All other specific embodiments obtained by those skilled in the art based on these specific embodiments without inventive effort are within the scope of protection of this invention.

[0044] To further understand the invention's content, features, and effects, the following specific embodiments are provided, along with accompanying drawings. Figure 1 Detailed explanation is as follows:

[0045] Example 1:

[0046] A knowledge graph-based method for real-time monitoring and intelligent operation and maintenance of turbine runners includes the following steps:

[0047] S1. Pressure sensors, vibration sensor arrays, and acoustic emission sensors are installed on the turbine runner;

[0048] Furthermore, the specific implementation method of step S1 includes the following steps:

[0049] S1.1. The vibration sensor array is arranged at equal angles at the blade roots and hub connection of the turbine runner, with the installation interval angles meeting the following requirements:

[0050]

[0051] in, This refers to the number of blades on the turbine runner. This represents the difference in installation angle between adjacent sensors;

[0052] S1.2. Pressure sensors are installed in the upper crown and lower circulation channel areas of the turbine runner, with a sampling frequency of... Set to 8-10 times the characteristic frequency of the turbine runner, satisfying:

[0053]

[0054] in, This refers to the number of blades on the turbine runner. The frequency corresponding to the rated speed;

[0055] S1.3. Acoustic emission sensors are placed in the following key areas of the turbine runner: the rounded corner area between the blade and the hub; the area near the middle of the blade near the inlet edge; the connection between the blade trailing edge and the cover plate; and the area near the connecting flange between the inner side of the hub and the main shaft.

[0056] S2. The signal acquisition module acquires the operating status parameters of the turbine runner in real time to obtain raw data;

[0057] Furthermore, a sliding window filtering algorithm is configured to process the original data. ,in The original signal, This is the filtered output. The length of the window;

[0058] S3. After preprocessing the raw data obtained in step S2, the graph construction module establishes a device model of the turbine runner and generates a knowledge graph of the turbine runner.

[0059] Furthermore, the specific implementation method of step S3 includes the following steps:

[0060] S3.1. Establish the equipment body model of the turbine runner. The mathematical expression is:

[0061]

[0062] in, , This represents a set of entities, including equipment entities, state entities, and process entities of the turbine runner. For the m-th specific entity; , Represents a set of relationships between entities; , For a collection of attributes, This is the kth specific attribute of the entity;

[0063] Furthermore, the graph neural network model is... ,in Represents a node In the The feature vector of the layer, The weight matrix is ​​trainable.

[0064] Furthermore, an attention mechanism is employed. Calculate the weights of multi-sensor data;

[0065] S3.2. An improved TransH algorithm is used to complete the knowledge of the turbine runner's equipment model, generating a knowledge graph of the turbine runner, expressed as:

[0066]

[0067] in, This is the overall vectorized representation of the knowledge graph of a water turbine runner. These represent the head entity vector and the tail entity vector, respectively. Let be the normal vector of the relation hyperplane. A translation vector specific to the relation.

[0068] Furthermore, a relationship strength decay function is set. ,in A knowledge forgetting factor; an event trigger mechanism is set up to detect parameter deviations. The map reconstruction process is initiated at the appropriate time;

[0069] S4. Based on the knowledge graph of the turbine runner obtained in step S2, the monitoring and analysis module analyzes the state of the turbine runner in real time and obtains the output of the monitoring and analysis module;

[0070] Furthermore, the specific implementation method of step S4 includes the following steps:

[0071] S4.1. Real-time monitoring data By embedding functions Incremental update of the overall vector of the knowledge graph transformed into that of a water turbine runner;

[0072] S4.2. Design a dynamic reasoning algorithm based on the matching degree of the overall vector of the knowledge graph of the turbine runner. Anomaly detection is performed, and the state of the turbine runner is analyzed in real time. The expression is:

[0073]

[0074] in, This is the current atlas fragment. For reference standard atlas.

[0075] Furthermore, wavelet packet decomposition algorithm is used to process the vibration signal:

[0076]

[0077] in, For the first Layer Wavelet packet coefficients of the node, This represents the energy characteristics of the corresponding frequency band. An adaptive alarm threshold is established based on a sliding time window.

[0078]

[0079] in, The mean of the window period characteristics. The standard deviation is denoted as .

[0080] S5. The intelligent operation and maintenance module generates maintenance strategy suggestions by matching historical cases through similarity calculation based on the output of the monitoring and analysis module.

[0081] Furthermore, in step S5, the method for generating maintenance strategy suggestions by matching historical cases through similarity calculation involves constructing a case reasoning engine, which uses similarity... The expression is used to calculate the matching historical case library:

[0082]

[0083] in, For the feature vector of the current problem, This is the feature vector of the i-th fault or operating condition case in the historical case library;

[0084] Design a strategy optimization loop based on the feedback matrix. The entity weight distribution of the knowledge graph is dynamically adjusted, expressed as:

[0085]

[0086] in, This represents the weight update value between the i-th entity and the j-th relation in the knowledge graph within the feedback matrix, used for reinforcement learning-based knowledge optimization mechanisms.

[0087] Furthermore, maintenance strategies are optimized through a reinforcement learning framework. ,in Indicates a maintenance action. This represents the device state vector.

[0088] Furthermore, a fault propagation model based on graph path analysis is established to analyze the fault propagation of abnormal events. Set of propagation paths within the time window ,in Indicates the first The propagation delay along each path; the development of a composite indicator evaluation system to comprehensively calculate the equipment health index. ,in For feature weights, This is the baseline parameter value.

[0089] Example 2:

[0090] A knowledge graph-based real-time monitoring and intelligent operation and maintenance system for turbine runners is implemented based on the knowledge graph-based real-time monitoring and intelligent operation and maintenance method for turbine runners described in Example 1. The system includes a signal acquisition module, a graph construction module, a monitoring and analysis module, and an intelligent operation and maintenance module, which are connected sequentially.

[0091] Furthermore, the aforementioned knowledge graph-based real-time monitoring and intelligent operation and maintenance system for turbine runners constructs a decision tree model based on Fault Mode and Effects Analysis (FMEA); integrates the bidirectional interaction function between the three-dimensional digital twin model and the knowledge graph topology; the system is configured with edge computing nodes and deploys a lightweight knowledge graph service framework, supporting 5ms-level real-time response.

[0092] Furthermore, an improved LSTM network is used to construct a lifetime prediction model:

[0093]

[0094]

[0095] in, These represent the forget gate, input gate, and output gate, respectively. Candidate memory units, The output is the hidden state. The model input is a multi-source feature vector. The output is the predicted remaining useful life. .

[0096] The 3D reconstruction of the wheel structure is achieved using WebGL technology, and the data mapping relationship satisfies:

[0097]

[0098] in, For three-dimensional coordinate points, For the corresponding monitoring parameters, This is a physics-based visualization mapping function. The alarm management terminal implements multi-level alarm classification and defines risk levels. ,in For severity, Let the probability of occurrence be... To increase the difficulty of detection, the system is configured with the following alarm levels: Level 1 (red): Immediate shutdown and inspection; Level 2 (orange): Planned maintenance; Level 3 (yellow): Continuous monitoring.

[0099] Furthermore, the system parameter configuration interface provides a function to set the knowledge graph update cycle, with the update frequency meeting the following requirements:

[0100]

[0101] in, To preset the minimum and maximum update intervals, This represents the current operating condition coefficient of the equipment. This is an adaptive adjustment factor.

[0102] It should be noted that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0103] Although this application has been described above with reference to specific embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of this application. In particular, as long as there is no structural conflict, the features in the specific embodiments disclosed in this application can be combined with each other in any way. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, this application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A knowledge graph-based real-time monitoring and intelligent operation and maintenance method for a hydraulic turbine runner, characterized in that, Comprising the following steps: S1. Arranging pressure sensors, vibration sensor arrays, acoustic emission sensors on the runner of the hydraulic turbine; S2. The signal acquisition module obtains the operating state parameters of the runner of the hydraulic turbine in real time to obtain original data; S3. The atlas construction module establishes the equipment ontology model of the runner of the hydraulic turbine after preprocessing the original data obtained in step S2 to generate the knowledge graph of the runner of the hydraulic turbine; S4. The monitoring and analysis module analyzes the state of the runner of the hydraulic turbine in real time based on the knowledge graph of the runner of the hydraulic turbine obtained in step S2 to obtain the output of the monitoring and analysis module; S5. The intelligent operation and maintenance module generates maintenance strategy suggestions by similarity calculation matching historical cases according to the output of the monitoring and analysis module.

2. The knowledge graph-based real-time monitoring and intelligent operation and maintenance method for a hydraulic turbine runner according to claim 1, characterized in that, The specific implementation method of step S1 comprises the following steps: S1.

1. The vibration sensor arrays are arranged at the blade roots and the hub connection of the runner of the hydraulic turbine in an equiangular distribution manner, and the installation interval angle satisfies: ； wherein, is the number of blades of the runner of the hydraulic turbine, is the difference in mounting angle of adjacent sensors; S1.

2. Pressure sensors are arranged in the upper crown and lower ring channel area of the water turbine runner, sampling frequency 8-10 times the characteristic frequency of the water turbine runner, satisfying: ； wherein, is the number of blades of the runner of the hydraulic turbine, is the rated rotational speed corresponding frequency; S1.

3. The acoustic emission sensors are arranged at the following key areas of the runner of the hydraulic turbine: the fillet area of the blade and the hub; the middle area of the blade close to the inlet edge; the connection part of the blade trailing edge and the cover plate; the inner side of the hub and the flange near the main shaft connection.

3. The knowledge graph-based real-time monitoring and intelligent operation and maintenance method for a hydraulic turbine runner according to claim 2, characterized in that, The specific implementation method of step S3 comprises the following steps: S3.

1. Establishing the equipment body model of the runner of the water turbine The mathematical expression is: ； wherein, , represents a set of entities, including equipment entities, state entities, process entities of a hydraulic turbine runner, is the mth specific entity; , represents a set of relationships between entities; , is a set of attributes, is the kth specific attribute of an entity; S3.

2. The improved TransH algorithm is used for knowledge completion of the equipment ontology model of the runner of the hydraulic turbine to generate the knowledge graph of the runner of the hydraulic turbine, and the expression is: ； wherein, is an overall vectorization representation of the knowledge graph of the runner of the hydraulic turbine, respectively represent the head entity vector and the tail entity vector, is a relation hyperplane normal vector, is a relation specific translation vector.

4. The knowledge graph-based real-time monitoring and intelligent operation and maintenance method for a hydraulic turbine runner according to claim 3, characterized in that, The specific implementation method of step S4 comprises the following steps: S4.

1. converting the real-time monitoring data by embedding functions transforming the incremental update of the knowledge graph vector of the turbine runner as a whole; S4.

2. Design a dynamic inference algorithm based on the matching degree of the overall vector of the water turbine runner's knowledge graph Abnormality detection is performed to analyze the state of the water turbine runner in real time, and the expression is: ； wherein, is the current atlas segment, is the reference standard atlas.

5. The knowledge graph-based real-time monitoring and intelligent operation and maintenance method for a hydraulic turbine runner according to claim 4, characterized in that, The method for generating the maintenance strategy suggestion by matching the historical cases through the similarity calculation in step S5 is to construct a case reasoning engine, and the historical case library is matched through the similarity The calculation is expressed as: ； wherein, is the current problem feature vector, is the feature vector of the i-th failure or operating condition case in the historical case base; Design a strategy optimization loop based on the feedback matrix Dynamically adjust the entity weight distribution of the knowledge graph, the expression is: ； wherein, is the weight update value of the i-th entity and j-th relation in the knowledge graph in the feedback matrix, used for the reinforcement learning type knowledge optimization mechanism.

6. A knowledge graph-based real-time monitoring and intelligent operation and maintenance system for water turbine runner, relying on the knowledge graph-based real-time monitoring and intelligent operation and maintenance method for water turbine runner according to any one of claims 1-5, characterized in that, It comprises a signal acquisition module, an atlas construction module, a monitoring and analysis module, and an intelligent operation and maintenance module, and the signal acquisition module, the atlas construction module, the monitoring and analysis module, and the intelligent operation and maintenance module are connected in sequence.

Images