Construction Method and Application Method of Vibration Fault Diagnosis System for Steam Turbine Generator Sets

Abstract

The present invention discloses a construction method and an application method for a vibration fault diagnosis system of a steam turbine generator set. The construction method includes: obtaining components and related states; analyzing the relationships of the states during the operation of the steam turbine generator set according to domain knowledge, as well as the relationships between components and states and the relationships between components when vibration faults occur, and generating rules based on the relationships; constructing an ontology model based on the components, states and rules; and performing self-optimization processing on the ontology model to obtain a vibration fault diagnosis system for the steam turbine generator set. The present invention constructs an ontology model for the vibration fault diagnosis system of the steam turbine generator set based on components, states and rules. In the construction of the rules, domain knowledge is combined with operation data, and the rules are constructed by analyzing the operation status and the fault generation process, thus solving the problem of difficult construction of the vibration fault system of the steam turbine generator set. In addition, the system has self-evaluation and correction capabilities, improving the reusability and diagnostic accuracy of the system.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and particularly relates to a construction method and an application method of a vibration fault diagnosis system for a steam turbine generator set. Background Art

[0002] Driven by Industry 4.0, the intelligent operation of power plants is gradually advancing, and the online vibration fault intelligent diagnosis system for steam turbine generator sets is one of the intelligent operations of power plants. Due to the large variety of equipment types in the steam turbine generator set itself, the large number of equipment involved in a single set of units, and the complex operation principle of the equipment itself, for the same unit, the same characteristics point to different fault causes, and for different units, the characteristics shown by the same fault cause are also different, making the composition of the vibration fault diagnosis system for steam turbine generator sets very complex, resulting in great difficulty in constructing the vibration fault diagnosis system for steam turbine generator sets.

[0003] In the prior art, the vibration fault diagnosis system for steam turbine generator sets is mainly constructed by machine learning or deep learning. This construction method ignores the operation logic of the vibration fault diagnosis system itself during the construction process, and the constructed system only extracts features from vibration data for diagnosis, resulting in low diagnostic accuracy of the constructed vibration fault diagnosis system for steam turbine generator sets. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to overcome the defects of great difficulty in constructing the vibration fault diagnosis system for steam turbine generator sets and low diagnostic accuracy in the prior art, and provide a construction method and an application method of a vibration fault diagnosis system for steam turbine generator sets.

[0005] The present invention solves the above technical problem by the following technical solutions:

[0006] According to the first aspect of the present invention, there is provided a construction method of a vibration fault diagnosis system for a steam turbine generator set, and the construction method of the vibration fault diagnosis system for the steam turbine generator set includes:

[0007] Obtain components of the steam turbine generator set; wherein, the components include the steam turbine, generator, excitation, and auxiliary system of the steam turbine generator set;

[0008] Obtain operation data from preset state measurement points of the steam turbine generator set, and extract the states related to the components from the operation data; wherein, the states include vibration state, thermal state, and electrical state;

[0009] Analyze the relationships between the states during the operation of the steam turbine generator set according to the domain knowledge of the steam turbine generator set and the operation data, as well as the relationships between the components and the related states and the relationships between the components when the steam turbine generator set has a vibration fault, and generate rules based on the relationships;

[0010] Based on the components, the states, and the rules, construct an ontology model of the vibration fault diagnosis system for the steam turbine generator set;

[0011] Apply the ontology model to the steam turbine generator set, and perform self-optimization processing on the ontology model according to the application effect of the ontology model after a period of time to obtain the vibration fault diagnosis system for the steam turbine generator set.

[0012] Preferably, the rules include operation rules, fault diagnosis rules, fault verification rules, and evaluation and learning rules. The step of generating rules based on the relationships includes:

[0013] Restore the operation analysis process of the steam turbine generator set according to the domain knowledge and the operation data, and generate the operation rules based on the relationships of the states in the operation analysis process;

[0014] Use the operation rules to detect whether the steam turbine generator set has abnormal states;

[0015] When the abnormal state occurs, restore the fault diagnosis process of the steam turbine generator set according to the domain knowledge and the abnormal state, and generate the fault diagnosis rules based on the relationships between the states and the components in the fault diagnosis process;

[0016] Use the fault diagnosis rules to find the faulty components and corresponding fault characteristics related to the abnormal state;

[0017] Restore the fault verification process of the steam turbine generator set according to the domain knowledge, the faulty components, and the fault characteristics, and generate the fault verification rules based on the relationships of the components in the fault verification process;

[0018] Use the fault verification rules to verify whether the faulty components are the fault causes of the abnormal state;

[0019] Restore the evaluation and learning process of the steam turbine generator set according to the operation data and the domain knowledge, and generate the evaluation and learning rules based on the relationships between the states and the components in the evaluation and learning process;

[0020] Use the evaluation and learning rules to evaluate the operation state of the steam turbine generator set to obtain an evaluation result, and then perform system self-learning or fault risk warning according to the evaluation result.

[0021] Preferably, the operation rules include standard operation rules and real-time operation rules. The step of generating the operation rules based on the relationships of the states in the operation analysis process includes:

[0022] Establish a standard operation library according to the domain knowledge, where the standard operation library includes a number of the standard operation rules; among them, the standard operation rules are represented as a number of standard states in an operation state;

[0023] Obtain the real-time state from the operation data, and generate the real-time operation rule based on the operation state of the standard operation rule and the real-time state; among them, the real-time operation rule is represented as a number of real-time states in an operation state.

[0024] Preferably, the rule further includes a fault rule, and the step of generating the fault diagnosis rule based on the relationship between the state and the component in the fault diagnosis process includes:

[0025] Establish a standard fault library according to the domain knowledge, where the standard fault library includes a number of the fault rules; among them, the fault rules are represented as a number of standard fault features under a faulty component;

[0026] Obtain the faulty component corresponding to the abnormal state; analyze the faulty component based on the fault rule, and find the fault feature in the faulty component that is consistent with the standard fault feature;

[0027] Generate the fault diagnosis rule according to the abnormal state, the faulty component and the first fault feature; among them, the fault diagnosis rule is represented as a number of faulty components and corresponding fault features in an abnormal state.

[0028] Preferably, the step of generating the fault verification rule based on the relationship between the components in the fault verification process includes:

[0029] Obtain the relevant features other than the fault feature in the faulty component, and verify whether the relevant features are consistent with the standard fault features based on the fault rule. If so, determine that the faulty component is the fault cause of the abnormal state;

[0030] Generate the fault verification rule according to the faulty component and the relevant features; among them, the fault verification rule is represented as a number of relevant features under a faulty component.

[0031] Preferably, the step of regularly evaluating the operation state of the steam turbine generator set by using the evaluation and learning rule includes:

[0032] Obtain a number of states to be evaluated from the operation data, and judge whether all the states to be evaluated meet the preset target threshold according to the domain knowledge. If all the states to be evaluated meet the target threshold, output the evaluation result that the operation state of the steam turbine generator set is stable;

[0033] If the to-be-evaluated state does not meet the target threshold, output the evaluation result that there is a fault risk in the steam turbine generator set.

[0034] Preferably, the step of performing system self-learning or fault risk warning according to the evaluation result includes:

[0035] If the evaluation result is that the operating state of the steam turbine generator set is stable, perform system self-learning to correct the standard operation library based on the operating state;

[0036] If the evaluation result is that there is a fault risk in the steam turbine generator set, perform a fault risk warning to output corresponding fault warning information based on the fault risk;

[0037] Evaluate the fault risk according to the domain knowledge to obtain an evaluation result, and then correct the standard fault library based on the evaluation result.

[0038] Preferably, after the step of constructing the ontology model of the vibration fault diagnosis system for the steam turbine generator set, it further includes:

[0039] Obtain an external reference model affecting the vibration fault diagnosis process of the steam turbine generator set based on the domain knowledge;

[0040] Integrate the external reference model into the ontology model by means of ontology integration.

[0041] Preferably, the step of integrating the external reference model into the ontology model by means of ontology integration includes:

[0042] Obtain the external components corresponding to the external state when the external state in the external reference model is the same as the state in the ontology model;

[0043] Add the external components to the components of the ontology model;

[0044] Obtain the external state corresponding to the external component when the external component in the external reference model is the same as the component in the ontology model;

[0045] Add the external state to the state of the ontology model;

[0046] Analyze the relationship between the external component and the state, and the relationship between the external state and the component according to the domain knowledge, and generate integration rules based on the relationship;

[0047] Add the integration rules to the rules of the ontology model.

[0048] According to a second aspect of the present invention, there is provided an application method of a vibration fault diagnosis system for a steam turbine generator set. The application method of the vibration fault diagnosis system for the steam turbine generator set includes the following steps:

[0049] Obtain operation data from preset state measurement points;

[0050] Analyze the operation data through the vibration fault diagnosis system of the steam turbine generator set to detect whether the steam turbine generator set has a vibration fault. If so, diagnose the vibration fault through the vibration fault diagnosis system of the steam turbine generator set and output a diagnosis result;

[0051] Evaluate the operation data through the vibration fault diagnosis system of the steam turbine generator set to determine whether the operation of the steam turbine generator set is stable. If so, correct the standard operation library based on the operation data;

[0052] If not, evaluate the fault risk based on the operation data, then correct the standard fault library based on the fault risk, and output corresponding fault warning information;

[0053] Wherein, the vibration fault diagnosis system of the steam turbine generator set is obtained by using the construction method of the vibration fault diagnosis system of the steam turbine generator set of the present invention.

[0054] The positive and progressive effects of the present invention are as follows:

[0055] Based on the concepts of ontology, namely components, states, and rules, the present invention constructs an ontology model of a vibration fault diagnosis system for a steam turbine generator set. When constructing the rules, based on domain knowledge and combined with on-site real-time data, analyze the operation status of the system to obtain operation rules, restore the fault generation process based on the operation logic to obtain fault diagnosis rules, verify the fault diagnosis using the fault generation mechanism to obtain fault verification rules, and restore the evaluation and learning process to obtain evaluation and learning rules, etc., which solves the problem of difficult construction of the vibration fault system for steam turbine generator sets. In addition, with the accumulation of the operation data of the system, the system is continuously corrected using the evaluation and learning rules of the system, and fault warnings are output in a timely manner, which not only improves the reusability and diagnostic accuracy of the system, but also effectively conducts online status monitoring, fault risk warning, and fault diagnosis of the steam turbine generator set, reducing the probability of faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a flowchart showing the construction method of the vibration fault diagnosis system for the steam turbine generator set according to Embodiment 1 of the present invention.

[0057] Figure 2 It is a flowchart showing step S13 of the construction method of the vibration fault diagnosis system for the steam turbine generator set according to Embodiment 1 of the present invention.

[0058] Figure 3 It is a schematic flowchart of step S131 in step S13 of the construction method of the vibration fault diagnosis system for a steam turbine generator set in Embodiment 1 of the present invention.

[0059] Figure 4 It is a schematic framework diagram of the operating condition analysis of the construction method of the vibration fault diagnosis system for a steam turbine generator set in Embodiment 1 of the present invention.

[0060] Figure 5 It is a schematic flowchart of step S133 in step S13 of the construction method of the vibration fault diagnosis system for a steam turbine generator set in Embodiment 1 of the present invention.

[0061] Figure 6 It is a schematic framework diagram of the fault diagnosis process of the construction method of the vibration fault diagnosis system for a steam turbine generator set in Embodiment 1 of the present invention.

[0062] Figure 7 It is a schematic flowchart of step S135 in step S13 of the construction method of the vibration fault diagnosis system for a steam turbine generator set in Embodiment 1 of the present invention.

[0063] Figure 8 It is a schematic framework diagram of the fault verification process of the construction method of the vibration fault diagnosis system for a steam turbine generator set in Embodiment 1 of the present invention.

[0064] Figure 9 It is a schematic flowchart of step S138 in step S13 of the construction method of the vibration fault diagnosis system for a steam turbine generator set in Embodiment 1 of the present invention.

[0065] Figure 10 It is a schematic flowchart of step S139 in step S13 of the construction method of the vibration fault diagnosis system for a steam turbine generator set in Embodiment 1 of the present invention.

[0066] Figure 11 It is a schematic framework diagram of the ontology model of the vibration fault diagnosis system for a steam turbine generator set in Embodiment 1 of the present invention.

[0067] Figure 12 It is a schematic flowchart of the construction method of the vibration fault diagnosis system for a steam turbine generator set in Embodiment 2 of the present invention.

[0068] Figure 13 It is a schematic flowchart of step S142 of the construction method of the vibration fault diagnosis system for a steam turbine generator set in Embodiment 2 of the present invention.

[0069] Figure 14 It is a schematic flowchart of the application method of the vibration fault diagnosis system for a steam turbine generator set in Embodiment 3 of the present invention. Detailed implementation manners

[0070] The present invention will be further described below by way of embodiments, but the present invention is not limited to the scope of the described embodiments for this reason.

[0071] Embodiment 1

[0072] This embodiment provides a construction method for a vibration fault diagnosis system of a steam turbine generator set. This construction method for a vibration fault diagnosis system of a steam turbine generator set can be applied to construct vibration fault diagnosis systems for steam turbine generator sets of various types of power plants. That is, the vibration fault diagnosis system constructed for the measured point data of one factory is dedicated. Under certain conditions, the ontology model of the vibration fault diagnosis system of the steam turbine generator set can be partially reused in another factory. It should be noted that this embodiment mainly uses the concept of ontology to manually construct the ontology of the system, and based on domain knowledge and operation data, models and analyzes the vibration fault diagnosis system of the steam turbine generator set, so as to construct the vibration fault diagnosis system of the steam turbine generator set.

[0073] Among them, the basic elements of the ontology include classes, attributes, and relationships. Among them, classes can also be called concepts. The ontology is used to describe the relationships between concepts and the relationships between attributes in concepts. As an optional implementation method for easy expression and understanding of modeling, in the vibration fault diagnosis system of the steam turbine generator set, components, states, and rules are used to represent concepts, attributes, and relationships, and an ontology model of the vibration fault diagnosis system of the steam turbine generator set is constructed based on components, states, and rules.

[0074] As Figure 1 shown, the construction method for this vibration fault diagnosis system of the steam turbine generator set includes the following steps:

[0075] S11. Obtain the components of the steam turbine generator set.

[0076] Among them, the components include the steam turbine, generator, excitation, and auxiliary system of the steam turbine generator set. As an optional implementation method, in each component, there are also several sub-components at different levels. For example, the steam turbine includes a stationary part and a rotor, etc., and the rotor includes a main shaft, an impeller, and moving blades, etc.

[0077] In this embodiment, the components of the steam turbine generator set are described in the form of a set, and the components of the steam turbine generator set are obtained according to the structure of the steam turbine generator set itself. Taking a certain type of steam turbine generator set as an example below, the components of the steam turbine generator set are obtained as follows:

[0078] Component = {steam turbine, generator, excitation, auxiliary system, other systems};

[0079] Steam turbine = {stationary part, rotor};

[0080] Stationary parts = {bedplate, cylinder, sliding key system, nozzle, diaphragm, diaphragm carrier ring, bearing, steam seal, fasteners, etc.}

[0081] Rotor = {main shaft, impeller, moving blade, steam seal, coupling}

[0082] Generator = {stator, rotor}

[0083] Stator = {bearing pedestal, seal gland, iron core, winding bar, stator outgoing lead, hydrogen, hydrogen cooler, elastic component}

[0084] Rotor = {main shaft, fan, wire slot, rotor winding, slot wedge, retaining ring, winding outgoing lead, coupling}

[0085] Excitation = {exciter, slip ring}

[0086] Auxiliary system = {heater, deaerator, extraction steam system, condenser, shaft seal system, drain system, vacuum system, feed pump set}

[0087] Other systems = {valve, oil system, cooling system, seal oil system}...

[0088] Of course, this embodiment is not limited to the above components and sub-components, and the definitions of components for different types of steam turbine generator sets are also different.

[0089] S12. Obtain the operation data from the preset state measurement points of the steam turbine generator set, and extract the state related to the components from the operation data.

[0090] Among them, the state includes vibration state, thermal state and electrical state. In order to ensure real-time tracking of the operation changes of the steam turbine generator set, a conventional steam turbine generator set operation system usually has 5000 - 30000 state measurement points, and the steam turbine generator set monitors its operation status through the operation data obtained from the state measurement points.

[0091] As an optional implementation manner, classify the state of the steam turbine generator set into vibration state category, thermal state category and electrical state category. Among them, the vibration state category includes key phase of the shafting, relative vibration of each bearing, pedestal vibration, axial displacement, cylinder expansion, differential expansion, eccentricity, etc.; the thermal state category includes temperature, pressure, flow rate, liquid level, etc. of steam, oil, cooling water, hydrogen, etc., temperature measurement of each metal component, valve position, opening degree, etc. of the valve; the electrical state category includes three-phase voltage, current, power, power factor, excitation voltage, excitation current, frequency, zero-sequence current, negative-sequence current of the generator, three-phase voltage, current, frequency, etc. on the transformer side.

[0092] In this embodiment, the state points to a unique state measurement point. The definition of the state includes a state value and state change characteristics. For example, in the vibration state of the shafting of a steam turbine generator unit, the vibration amplitude of the shafting is the state value, and the occurrence of a jump in the vibration of the shafting is the state change characteristic. As an alternative embodiment, the state can point to different components, that is, the states within the components and the states between related components, such as the steam circuit and the electromagnetic circuit.

[0093] S13. Analyze the relationships of the states during the operation of the steam turbine generator unit based on the domain knowledge and operation data of the steam turbine generator unit, as well as the relationships between components and related states and the relationships between components when vibration faults occur in the steam turbine generator unit, and generate rules based on the relationships.

[0094] In this embodiment, the operation analysis process of the steam turbine generator unit refers to the process of jointly completing the conversion of thermal energy - mechanical energy - electrical energy by the interaction of the steam turbine, generator, auxiliary equipment and other equipment. Vibration fault diagnosis is based on the dynamic state and domain knowledge of the steam turbine generator unit, restoring the operation analysis process, and confirming whether each component and its sub-components in the steam turbine generator unit are in their proper positions during operation through rules. The rules therein represent a logical description of the relationships and state relationships of each system or subsystem based on the working principle of the system.

[0095] As an alternative embodiment, the rules are mainly constructed and described through the components and states of the steam turbine generator unit. According to the fault generation mechanism of the steam turbine generator unit, the construction of the rules includes operation state analysis, fault diagnosis process, fault verification process, evaluation and learning process, etc. The defined rules include operation rules, fault diagnosis rules, fault verification rules and evaluation and learning rules.

[0096] See Figure 2 , step S13 specifically includes the following steps:

[0097] S131. Restore the operation analysis process of the steam turbine generator unit based on the domain knowledge and operation data, and generate operation rules based on the relationships of the states during the operation analysis process.

[0098] Among them, the operation rules refer to a set of some states of the operation analysis process of the steam turbine generator unit. The states include state values and state change characteristics. As an alternative embodiment, the operation rules are described in the form of a set, and the operation rules = {under a certain operation state: state value set} + {under a certain operation state: state change characteristic set}, the state value set = {state value 1,..., state value n}, the state change characteristic set = {state change characteristic 1,..., state change characteristic n}.

[0099] In this embodiment, the operation rules include standard operation rules and real-time operation rules. The standard operation rules are a state library formed based on domain knowledge, and the real-time operation rules are a state library generated by analyzing operation data according to the state library definition in the standard state rules.

[0100] See Figure 3 , step S131 specifically includes the following steps:

[0101] S1311. Establish a standard operation library according to domain knowledge. The standard operation library includes a number of standard operation rules.

[0102] Among them, the standard operation rules are expressed as a number of standard states in an operation state. In this embodiment, the standard operation library is the standard operation rule library, that is, a set of a series of standard operation rules. The standard operation rules are a set of standard states in an operation state. Among them, the standard state refers to the state of the components of the steam turbine generator set when it is theoretically operating stably through domain knowledge analysis.

[0103] As an optional implementation manner, the standard operation rules = {operation state: standard state set}, and the standard operation library = {operation state 1: standard state set 1, operation state 2: standard state set 2,... operation state n : standard state set n}.

[0104] S1312. Obtain the real-time state from the operation data, and generate real-time operation rules based on the operation state and real-time state of the standard operation rules.

[0105] Among them, the real-time operation rules are expressed as a number of real-time states in an operation state. In this embodiment, the real-time operation rules are a set of real-time states in an operation state. Among them, the real-time state refers to the state of the components of the steam turbine generator set when it is actually operating through the operation data analysis of the state measurement points. As an optional implementation manner, the real-time operation rules = {operation state: real-time state set}.

[0106] S132. Use the operation rules to detect whether the steam turbine generator set has an abnormal state. If so, execute step S133.

[0107] As an optional implementation manner, the operation analysis process is a comparison process between the real-time operation rules and the standard operation rules. Among them, both the real-time operation rules and the standard operation rules perform the same analysis and calculation for a specific operation state. By comparing these two rules, if the deviation between the real-time operation rules and the standard operation rules is large, an abnormal state set is detected. Such as Figure 4As shown, the result of running the analysis process points to a set of abnormal states. As an alternative implementation, all real-time states in the same running state are sequentially determined to see if they are within the range of the standard state. If not, the corresponding real-time state is determined to be an abnormal state, and then fault diagnosis is performed based on the abnormal state.

[0108] S133. Restore the fault diagnosis process of the steam turbine generator set according to domain knowledge and abnormal states, and generate fault diagnosis rules based on the relationship between states and components during the fault diagnosis process.

[0109] As an alternative implementation, the fault diagnosis process refers to a series of specific analysis processes carried out according to the set of abnormal states to find the corresponding set of relevant components, and to analyze a certain abnormal state of a component or sub-component in the component set. A number of fault characteristics generated from this component set are combined into a fault characteristic set, and this fault characteristic set is compared with the defined standard fault library to analyze possible fault causes, that is, to find the faulty component or sub-component.

[0110] See Figure 5 , step S133 specifically includes the following steps:

[0111] S1331. Establish a standard fault library according to domain knowledge. The standard fault library includes a number of fault rules.

[0112] Among them, the fault rule is expressed as a number of standard fault characteristics under a faulty component. In this embodiment, the standard fault library is a set of a series of fault rules, and the fault rule is a set of standard fault characteristics under a faulty component. Among them, the standard fault characteristic refers to the characteristic of the component of the steam turbine generator set when an abnormal fault occurs through domain knowledge analysis.

[0113] As an alternative implementation, the characteristics are classified according to operating state characteristics, position characteristics, vibration characteristics, and other state characteristics. Among them, the operating state characteristics refer to the set of operating states during the operation analysis process; the position characteristics refer to the position attributes of the faulty component; the vibration characteristics refer to the vibration information of the faulty component, and the vibration characteristics include vibration signal type, multiple frequency amplitude, multiple frequency angle, shaft center position characteristics, shaft center trajectory characteristics, and vibration abnormal change characteristics, etc.; the other state characteristics refer to the set of all state change characteristic values that do not belong to the vibration characteristics.

[0114] As an alternative implementation, the fault rule = {faulty component: standard fault characteristic set}, and the standard fault library = {faulty component 1: standard fault characteristic set 1, faulty component 2: standard fault characteristic set 2,... faulty component n : standard fault characteristic set n}, it should be noted that the faulty component can be a certain component or sub-component in the component set.

[0115] S1332. Obtain the faulty component corresponding to the abnormal state.

[0116] Since in this embodiment, the state points to a unique state measurement point, the corresponding state measurement point can be queried through the abnormal state, and the relevant component set, that is, the faulty component, can be obtained through the state measurement point.

[0117] S1333. Analyze the faulty components based on the fault rules to find the fault features in the faulty components that are consistent with the standard fault features.

[0118] As an optional implementation manner, match the features of the faulty components with the standard fault features in the standard fault library. If there are features in the faulty components that are consistent with the standard fault features, then regard these features as fault features, and regard the corresponding faulty components as the possible fault causes. It should be noted that the above-mentioned fault features are not fixed, but have a certain degree of randomness. This is because the composition of the steam turbine generator set is very complex. Considering the differences in on-site operating conditions, environment, and operation management, for the same unit, the same feature may point to different fault causes; for different units, the same fault cause may have different manifestations of features. Therefore, even for the same fault, different fault features may be obtained in different fault diagnosis processes.

[0119] S1334. Generate a fault diagnosis rule according to the abnormal state, faulty components, and fault features.

[0120] Among them, the fault diagnosis rule is expressed as several faulty components and corresponding fault features under an abnormal state.

[0121] As an optional implementation manner, describe the fault diagnosis rule in the form of a set, and obtain the fault diagnosis rule = {abnormal state: faulty component set} + {faulty component: fault feature}...

[0122] S134. Use the fault diagnosis rule to find the faulty components and corresponding fault features related to the abnormal state.

[0123] As an optional implementation manner, the fault diagnosis process is a process of analyzing the faulty components and corresponding fault features according to the set of abnormal states and the standard fault library. As Figure 6 shown, the result of the fault diagnosis process is the fault cause set, which points to an abnormal state of a faulty component or sub-component. The faulty component set is a list of faulty components that may exist under the abnormal state, and the fault cause set is the fault features and corresponding faulty components obtained after matching the faulty component set with the standard fault library.

[0124] S135. Restore the fault verification process of the steam turbine generator set based on domain knowledge, faulty components, and fault characteristics, and generate fault verification rules based on the relationships between components during the fault verification process.

[0125] As an optional implementation manner, the fault verification process refers to, according to the obtained list of faulty components, further comparing the remaining characteristics with the standard fault library by analyzing the remaining characteristics under each faulty component one by one to verify whether this faulty component or sub-faulty component is valid.

[0126] See Figure 7 , step S135 specifically includes the following steps:

[0127] S1351. Obtain the relevant characteristics of the faulty component except for the first fault characteristic.

[0128] Among them, the fault characteristic refers to the characteristic directly corresponding to the abnormal state, and the relevant characteristic refers to the characteristic that causes other abnormalities in the faulty component when the fault characteristic holds. As an optional implementation manner, some relevant characteristics related to the fault characteristic can be obtained for verification, or all relevant characteristics except the fault characteristic can be obtained for verification.

[0129] S1352. Based on the fault rule, verify whether the relevant characteristic is consistent with the standard fault characteristic. If so, execute step S1353.

[0130] S1353. Determine the fault cause for the faulty component to be in the abnormal state.

[0131] As an optional implementation manner, after determining the fault cause for a certain faulty component to be in the abnormal state, output the faulty component and the relevant fault cause as the diagnosis result.

[0132] S1354. Generate a fault verification rule according to the faulty component and the relevant characteristics; among them, the fault verification rule is expressed as several relevant characteristics under a faulty component.

[0133] As an optional implementation manner, describe the fault verification rule in the form of a set, and obtain the fault diagnosis feature = {faulty component: relevant characteristic}...

[0134] As an optional implementation manner, the relevant characteristics can be converged to the fault diagnosis rule, that is, directly analyze the relevant characteristics during the fault diagnosis process.

[0135] S136. Use the fault verification rule to verify whether the faulty component is the fault cause of the abnormal state.

[0136] As an optional implementation manner, the fault verification process is a process of verifying whether a faulty component is valid according to the remaining characteristics under a single component and the standard fault library, such asFigure 8 As shown, the result of the fault verification process is the cause of the fault. Assume that the set of fault causes includes faulty component X and faulty component Y. Respectively, match the remaining features (i.e., relevant features) other than the fault features under faulty component X and faulty component Y with the standard fault library. If there is a fault in the relevant features, determine that the faulty component corresponding to the relevant features is the cause of the fault. Refer to Figure 8 , there is a fault in the relevant features of faulty component X, so faulty component X is the cause of the fault.

[0137] Here is an example. For instance, when the vibration fluctuation of a certain measuring point of the turbogenerator shafting is abnormally large, through the fault diagnosis process, it is diagnosed that the rotor mass may be unbalanced, but it may also be the measurement error of this measuring point. Then, the shaft vibration and bearing vibration of other measuring points can be verified to determine whether the change amplitude of the shaft vibration or bearing vibration is abnormal. If it is abnormal, confirm that the rotor is the cause of the fault and perform maintenance. If the data of the remaining measuring points is normal, perform maintenance on this measuring point. Of course, in the actual process of the occurrence of turbogenerator vibration faults, the diagnostic situation is much more complex. Therefore, it is necessary to further determine the cause of the fault through the fault verification process after the fault diagnosis process.

[0138] S137. Restore the evaluation and learning process of the turbogenerator set according to the operation data and domain knowledge, and generate the evaluation and learning rules based on the relationship between the states and components in the evaluation and learning process.

[0139] Among them, the evaluation and learning process is equivalent to regularly "examining" the fault diagnosis system of the turbogenerator set, using the operation data of a certain time period to evaluate the operation state of the system, evaluating the stability and risks of the system. In the stable case, the standard operation library is corrected through self-learning. In the case of risks, fault warning information is output, and by further analyzing the risks, the standard fault library is learned and corrected, etc.

[0140] S138. Regularly evaluate the operation state of the turbogenerator set using the evaluation and learning rules to obtain an evaluation result;

[0141] Since the parameters of different types of turbogenerator sets are different, and when the accumulation of original data is small, the standard operation library and standard fault library constructed through domain knowledge and operation data may be insufficient, resulting in low applicability and diagnostic accuracy of the ontology model of the turbogenerator set vibration fault diagnosis system. Therefore, the turbogenerator set vibration fault diagnosis system also includes an evaluation mechanism and a learning mechanism to achieve system cycle improvement.

[0142] Refer to Figure 9 , step S138 specifically includes the following steps:

[0143] S1381. Obtain several states to be evaluated from the operation data.

[0144] In this embodiment, the system evaluates based on the operation data of the steam turbine generator set for a certain period of time, that is, regularly evaluates the operation state of the system, evaluates the stability and risks of the system. Specifically, every other cycle, operation data is obtained from the state measurement points, and several states in each operation state within this cycle are obtained, and these states are used as the states to be evaluated.

[0145] S1382. Determine whether all the states to be evaluated meet the preset target threshold according to the domain knowledge. If all the states to be evaluated meet the target threshold, execute step S1383; if there is a state to be evaluated that does not meet the target threshold, execute step S1384.

[0146] As an optional implementation manner, the evaluation mainly includes vibration state evaluation, bearing stability evaluation, and overall system stability evaluation. Among them, the vibration state evaluation mainly includes transient and steady-state evaluations of the vibration state in each operation state; the bearing stability evaluation mainly includes an evaluation of the bearing vibration state combined with the thermal parameters of the bearing bush; the overall system stability evaluation mainly includes an evaluation of the system's anti-interference ability.

[0147] As an optional implementation manner, for a certain operation state within a certain period of time, sequentially determine whether several states to be evaluated in this operation state meet the threshold range (i.e., the target threshold) set by the domain knowledge. If all the states to be evaluated are within the threshold range, determine that this operation state is stable and execute step S1383; if there is a state to be evaluated that is not within the threshold range, determine that this operation state is unstable and execute step S1384.

[0148] S1383. Output the evaluation result that the operation state of the steam turbine generator set is stable.

[0149] S1384. Output the evaluation result that there is a fault risk in the steam turbine generator set.

[0150] S139. Perform system self-learning or fault risk warning according to the evaluation result.

[0151] See Figure 10 , if the evaluation result is that the operation state of the steam turbine generator set is stable, execute step S1391; if the evaluation result is that there is a fault risk in the steam turbine generator set, execute step S1392. Specifically, step S139 includes the following steps:

[0152] S1391. Perform system self-learning to correct the standard operation library based on the operation state.

[0153] As an optional implementation manner, automatically record the state in this operation state as a new standard state, update it to the standard operation library, and then correct the operation rules based on the standard operation library.

[0154] In this embodiment, automatic learning or semi-automatic learning can be performed according to the learning mechanism. Among them, automatic learning is to optimize the system based on the evaluation mechanism and learning mechanism of domain knowledge, such as correcting or updating relevant calculation parameters, and correcting or converging the state value set and state change feature set in each rule. Semi-automatic learning is to manually adjust the system. For example, in the initial stage of system operation when the accumulation of original state records is small, the parameters learned by similar units are used, and the parameters are appropriately adjusted through manual intervention in combination with the on-site operation status. As an alternative implementation, semi-automatic learning can also be to manually intervene to increase or decrease certain rule bases during the process of system operation and maintenance; or manually intervene to select eligible time periods to train and learn certain rules to force the convergence of certain states, etc.

[0155] S1392. Perform fault risk early warning to output corresponding fault warning information based on the fault risk.

[0156] S1393. Evaluate the fault risk according to domain knowledge to obtain an evaluation result.

[0157] In this embodiment, the process of evaluating the fault risk is the same as that of the fault diagnosis process and the fault verification process, except that the criteria are different. The fault risk is a risk prompt before a fault occurs. As an alternative implementation, obtain a set of risk states that do not meet the target threshold, analyze this set of risk states, and confirm the risk components and corresponding risk characteristics (i.e., the evaluation result).

[0158] S1394. Modify the standard fault library based on the evaluation result.

[0159] As an alternative implementation, when evaluating the fault risk, if the risk state always points to certain components or sub-components during multiple evaluation processes, the above risk state and related component sets can be converged, and the risk components and corresponding risk characteristics are updated to the standard fault library, which is beneficial to narrowing the scope of fault causes during the fault diagnosis process.

[0160] This embodiment continuously corrects the rules of the ontology model by using the evaluation and learning rules, so that the ontology model of the steam turbine generator unit vibration fault diagnosis system can be adapted to different types and parameters of steam turbine generator units, thereby obtaining a dedicated steam turbine generator unit vibration fault diagnosis system; in addition, when a fault risk is evaluated, a fault warning information is output in a timely manner, which can reduce the probability of major faults occurring.

[0161] S14. Based on components, states, and rules, construct an ontology model of the steam turbine generator unit vibration fault diagnosis system.

[0162] As an alternative implementation, such as Figure 11As shown in the figure, the ontology model of the vibration fault diagnosis system for a steam turbine generator set is presented. In the ontology model of the vibration fault diagnosis system for a steam turbine generator set, components and states are defined. In the construction of rules, the detailed construction processes of several types of rules are mainly included, such as standard operation rules, real-time operation rules, fault rules, fault diagnosis rules, fault verification rules, and evaluation and learning rules. And these types of rules are described based on components and states.

[0163] As an optional implementation manner, the ontology model also includes the relationships of these types of rules. That is, the abnormal state set is obtained by comparing the real-time operation rules and the standard operation rules. Based on the abnormal state set and the fault rules, the fault diagnosis rules are obtained. Based on the fault diagnosis rules, the fault cause set is obtained. Based on the fault rules and the fault cause set, the fault verification rules are obtained. The evaluation and learning rules correct the standard operation library when the system is stable, and then update the standard operation rules. When there is a fault risk in the system, the standard fault library is corrected, and then the fault diagnosis rules and the fault verification rules are updated.

[0164] S15. Apply the ontology model to the steam turbine generator set, and perform self-optimization processing on the ontology model according to the application effect after a period of time to obtain the vibration fault diagnosis system for the steam turbine generator set.

[0165] This embodiment constructs the ontology model of the vibration fault diagnosis system for a steam turbine generator set based on the concepts of ontology, namely components, states, and rules. When constructing the rules, based on domain knowledge and combined with on-site operation data, the operation rules are obtained by analyzing the operation status of the system. The fault diagnosis rules are obtained by restoring the fault generation process based on the operation logic. The fault verification rules are obtained by verifying the fault diagnosis using the fault generation mechanism. And the evaluation and learning rules are obtained by restoring the evaluation and learning process, etc. This solves the problem of difficult construction of the vibration fault system for a steam turbine generator set. In addition, with the accumulation of the system's operation data, the system is continuously corrected using the evaluation and learning rules of the system, and fault warnings are output in a timely manner, which not only reduces the probability of faults but also improves the reusability and diagnostic accuracy of the system.

[0166] Embodiment 2

[0167] This embodiment provides a construction method for a vibration fault diagnosis system for a steam turbine generator set. This construction method for the vibration fault diagnosis system for a steam turbine generator set is a further improvement of Embodiment 1. As Figure 12 shown, after step S14, step S141 is further executed. Specifically, it includes the following steps:

[0168] S141. Obtain an external reference model that affects the vibration fault diagnosis process of the steam turbine generator set based on domain knowledge.

[0169] As an alternative implementation, from the perspective of systems engineering, considering the impact on the system outside the vibration fault diagnosis system of the steam turbine generator unit, the environment and the power grid are introduced, and the environment and the power grid are integrated into the ontology model of the vibration fault diagnosis system of the steam turbine generator unit as an external reference model.

[0170] As an alternative implementation, the environment and the power grid can be a real-time model or a model generated using historical data.

[0171] S142. Integrate the external reference model into the ontology model by means of ontology integration.

[0172] See Figure 13 , step S142 specifically includes the following steps:

[0173] S1421. Obtain the external components corresponding to the external state when the external state in the external reference model is the same as the state in the ontology model.

[0174] As an alternative implementation, the components related to the ontology model of the vibration fault diagnosis system of the steam turbine generator unit in the environment reference model include the oil system, the cooling system, the generator, etc.; the components related to the ontology model of the vibration fault diagnosis system of the steam turbine generator unit in the power grid reference model include the generator, the steam turbine, the shafting, etc.

[0175] S1422. Add the external components to the components in the ontology model.

[0176] S1423. Obtain the external state corresponding to the external component when the external component in the external reference model is the same as the component in the ontology model.

[0177] As an alternative implementation, the states related to the ontology model of the vibration fault diagnosis system of the steam turbine generator unit in the environment reference model include temperature, humidity, air pressure, region, etc.; the states related to the ontology model of the vibration fault diagnosis system of the steam turbine generator unit in the power grid reference model include grid load and grid quality, etc.

[0178] S1424. Add the external state to the states in the ontology model.

[0179] S1425. Analyze the relationships between the external components and states, and between the external states and components according to domain knowledge, and generate integration rules based on the relationships.

[0180] S1426. Add the integration rules to the rules in the ontology model.

[0181] In this embodiment, considering the impact on the system outside the vibration fault diagnosis system of the steam turbine generator unit, data of the environment and the power grid are introduced to participate in the calculation and analysis of this system, improving the ontology model of the vibration fault diagnosis system of the steam turbine generator unit, and further enhancing the accuracy of the vibration fault diagnosis system of the steam turbine generator unit in fault diagnosis.

[0182] Embodiment 3

[0183] This embodiment provides an application method for a vibration fault diagnosis system of a steam turbine generator unit. As Figure 14 shown, the application method for the vibration fault diagnosis system of the steam turbine generator unit includes the following steps:

[0184] S21. Obtain operation data from preset state measurement points.

[0185] S22. Analyze the operation data through the vibration fault diagnosis system of the steam turbine generator unit to detect whether a vibration fault occurs. If so, execute step S23.

[0186] Among them, the vibration fault diagnosis system of the steam turbine generator unit is obtained by using the construction method of the vibration fault diagnosis system of the steam turbine generator unit in Embodiment 1 or Embodiment 2.

[0187] S23. Diagnose the vibration fault through the vibration fault diagnosis system of the steam turbine generator unit.

[0188] S24. Output the diagnosis result.

[0189] S25. Evaluate the operation data through the vibration fault diagnosis system of the steam turbine generator unit to determine whether the operation of the steam turbine generator unit is stable. If so, execute step S26; if not, execute step S17.

[0190] S26. Modify the standard operation library based on the operation data.

[0191] S27. Evaluate the fault risk based on the operation data.

[0192] S28. Modify the standard fault library based on the fault risk.

[0193] S29. Output the corresponding fault warning information.

[0194] In this embodiment, the vibration fault diagnosis system of the steam turbine generator set for fault early warning and fault diagnosis is obtained by the construction method of the vibration fault diagnosis system of the steam turbine generator set in Embodiment 1 or Embodiment 2. Therefore, when the vibration fault diagnosis system of the steam turbine generator set is applied to specific fault early warning and fault diagnosis, it can effectively perform online condition monitoring, fault risk early warning and fault diagnosis on the steam turbine generator set, reduce the probability of major faults occurring, and provide a reference for the maintenance and repair of the system, improving the on-site operation stability and safety of the steam turbine generator set in the power plant.

[0195] Although the specific embodiments of the present invention have been described above, those skilled in the art should understand that this is only an example. The protection scope of the present invention is defined by the appended claims. Without departing from the principles and essence of the present invention, those skilled in the art can make various changes or modifications to these embodiments, but these changes and modifications all fall within the protection scope of the present invention.

Claims

1. A method for constructing a vibration fault diagnosis system for a steam turbine generator set, characterized in that, The construction method of the vibration fault diagnosis system for the steam turbine generator set includes: Obtaining the components of the steam turbine generator set; wherein, the components include the steam turbine, generator, excitation and auxiliary systems of the steam turbine generator set; Obtaining operation data from the preset state measurement points of the steam turbine generator set, and extracting the states related to the components from the operation data; wherein, the states include vibration state, thermal engineering state and electrical state; Analyzing the relationship between the states during the operation of the steam turbine generator set according to the domain knowledge of the steam turbine generator set and the operation data, as well as the relationships between the components and the relevant states and the relationships between the components when the steam turbine generator set has a vibration fault, and generating rules based on the relationships; Based on the components, the states and the rules, constructing an ontology model of the vibration fault diagnosis system for the steam turbine generator set; Applying the ontology model to the steam turbine generator set, and performing self-optimization processing on the ontology model according to the application effect of the ontology model after a period of time to obtain the vibration fault diagnosis system for the steam turbine generator set; The rules include operation rules, fault diagnosis rules, fault verification rules and evaluation and learning rules, and the step of generating rules based on the relationships includes: Restoring the operation analysis process of the steam turbine generator set according to the domain knowledge and operation data, and generating the operation rules based on the relationships between the states in the operation analysis process; Using the operation rules to detect whether the steam turbine generator set has an abnormal state; When the abnormal state occurs, restoring the fault diagnosis process of the steam turbine generator set according to the domain knowledge and the abnormal state, and generating the fault diagnosis rules based on the relationships between the states and the components in the fault diagnosis process; Using the fault diagnosis rules to find the fault components and the corresponding fault characteristics related to the abnormal state; Restoring the fault verification process of the steam turbine generator set according to the domain knowledge, the fault components and the fault characteristics, and generating the fault verification rules based on the relationships between the components in the fault verification process; Using the fault verification rules to verify whether the fault components are the fault causes of the abnormal state Restoring the evaluation and learning process of the steam turbine generator set according to the operation data and the domain knowledge, and generating the evaluation and learning rules based on the relationships between the states and the components in the evaluation and learning process; Using the evaluation and learning rules to regularly evaluate the operation state of the steam turbine generator set to obtain an evaluation result, and then performing system self-learning or fault risk warning according to the evaluation result.

2. The method for constructing a vibration fault diagnosis system for a steam turbine generator set according to claim 1, wherein The operation rules include standard operation rules and real-time operation rules, and the step of generating the operation rules based on the relationships between the states in the operation analysis process includes: Establishing a standard operation library according to the domain knowledge, and the standard operation library includes a number of the standard operation rules; wherein, the standard operation rules are expressed as a number of standard states in an operation state; Obtain the real-time status from the operation data, and generate the real-time operation rule based on the operation status of the standard operation rule and the real-time status; wherein, the real-time operation rule is represented as several real-time statuses in an operation state.

3. The construction method of the vibration fault diagnosis system for a steam turbine generator set according to claim 1, characterized in that, The rule further includes a fault rule, and the step of generating the fault diagnosis rule based on the relationship between the status and the component in the fault diagnosis process includes: Establish a standard fault library according to the domain knowledge, and the standard fault library includes several of the fault rules; wherein, the fault rule is represented as several standard fault characteristics under a faulty component. Obtain the faulty component corresponding to the abnormal status; analyze the faulty component based on the fault rule, and find the fault characteristics in the faulty component that are consistent with the standard fault characteristics. Generate the fault diagnosis rule according to the abnormal status, the faulty component and the fault characteristics; wherein, the fault diagnosis rule is represented as several faulty components and corresponding fault characteristics in an abnormal state.

4. The construction method of the vibration fault diagnosis system for a steam turbine generator set according to claim 3, characterized in that, The step of generating the fault verification rule based on the relationship between the components in the fault verification process includes: Obtain the relevant characteristics of the faulty component other than the fault characteristics, and verify whether the relevant characteristics are consistent with the standard fault characteristics based on the fault rule. If so, determine that the faulty component is the cause of the abnormal status. Generate the fault verification rule according to the faulty component and the relevant characteristics; wherein, the fault verification rule is represented as several relevant characteristics under a faulty component.

5. The construction method of the vibration fault diagnosis system for a steam turbine generator set according to claim 1, characterized in that, The step of periodically evaluating the operation status of the steam turbine generator set by using the evaluation and learning rule includes: Obtain several statuses to be evaluated from the operation data, and judge whether all the statuses to be evaluated meet the preset target threshold according to the domain knowledge. If all the statuses to be evaluated meet the target threshold, output the evaluation result that the operation status of the steam turbine generator set is stable. If there is a status to be evaluated that does not meet the target threshold, output the evaluation result that there is a fault risk in the steam turbine generator set.

6. The construction method of the vibration fault diagnosis system for a steam turbine generator set according to claim 1, characterized in that The step of performing system self-learning or fault risk warning according to the evaluation result includes: If the evaluation result is that the operation status of the steam turbine generator set is stable, perform system self-learning to correct the standard operation library based on the operation status. If the evaluation result is that there is a fault risk in the steam turbine generator set, perform a fault risk warning to output the corresponding fault warning information based on the fault risk. Evaluate the fault risk according to the domain knowledge to obtain an evaluation result, and then correct the standard fault library based on the evaluation result.

7. The method for constructing a vibration fault diagnosis system of a steam turbine generator set according to claim 1, characterized in that After the step of constructing the ontology model of the vibration fault diagnosis system for the steam turbine generator set, it further includes: Obtain an external reference model that affects the vibration fault diagnosis process of the steam turbine generator set according to the domain knowledge. Integrate the external reference model into the ontology model by using the ontology integration method.

8. The construction method of the vibration fault diagnosis system for a steam turbine generator set according to claim 7, characterized in that The step of integrating the external reference model into the ontology model by using the ontology integration method includes: Obtain the external components corresponding to the external state when the external state in the external reference model is the same as the state of the ontology model; Add the external components to the components of the ontology model; Obtain the external state corresponding to the external components when the external components in the external reference model are the same as the components of the ontology model; Add the external state to the states of the ontology model; Analyze the relationship between the external components and the states, and the relationship between the external states and the components according to the domain knowledge, and generate integration rules based on the relationships; Add the integration rules to the rules of the ontology model.

9. An application method of a vibration fault diagnosis system for a steam turbine generator set, characterized in that, The application method of the steam turbine generator set vibration fault diagnosis system includes the following steps: Obtain operation data from preset state measurement points; Analyze the operation data through the steam turbine generator set vibration fault diagnosis system to detect whether the steam turbine generator set has a vibration fault. If so, diagnose the vibration fault through the steam turbine generator set vibration fault diagnosis system and output a diagnosis result; Evaluate the operation data through the steam turbine generator set vibration fault diagnosis system to judge whether the operation of the steam turbine generator set is stable. If so, correct the standard operation library based on the operation data; If not, evaluate the fault risk based on the operation data, then correct the standard fault library based on the fault risk, and output the corresponding fault warning information; Wherein, the steam turbine generator set vibration fault diagnosis system is obtained by using the construction method of the steam turbine generator set vibration fault diagnosis system according to any one of claims 1-8.

Images